This is done automatically with `gulp lint --fix` and the following
manual changes:
```diff
diff --git a/src/core/function.js b/src/core/function.js
index 878001057..b7e3e6ccf 100644
--- a/src/core/function.js
+++ b/src/core/function.js
@@ -131,7 +131,7 @@ function toNumberArray(arr) {
return arr;
}
-var PDFFunction = (function PDFFunctionClosure() {
+const PDFFunction = (function PDFFunctionClosure() {
const CONSTRUCT_SAMPLED = 0;
const CONSTRUCT_INTERPOLATED = 2;
const CONSTRUCT_STICHED = 3;
@@ -484,7 +484,9 @@ var PDFFunction = (function PDFFunctionClosure() {
// clip to domain
const v = clip(src[srcOffset], domain[0], domain[1]);
// calculate which bound the value is in
- for (var i = 0, ii = bounds.length; i < ii; ++i) {
+ const length = bounds.length;
+ let i;
+ for (i = 0; i < length; ++i) {
if (v < bounds[i]) {
break;
}
@@ -673,23 +675,21 @@ const PostScriptStack = (function PostScriptStackClosure() {
roll(n, p) {
const stack = this.stack;
const l = stack.length - n;
- let r = stack.length - 1,
- c = l + (p - Math.floor(p / n) * n),
- i,
- j,
- t;
- for (i = l, j = r; i < j; i++, j--) {
- t = stack[i];
+ const r = stack.length - 1;
+ const c = l + (p - Math.floor(p / n) * n);
+
+ for (let i = l, j = r; i < j; i++, j--) {
+ const t = stack[i];
stack[i] = stack[j];
stack[j] = t;
}
- for (i = l, j = c - 1; i < j; i++, j--) {
- t = stack[i];
+ for (let i = l, j = c - 1; i < j; i++, j--) {
+ const t = stack[i];
stack[i] = stack[j];
stack[j] = t;
}
- for (i = c, j = r; i < j; i++, j--) {
- t = stack[i];
+ for (let i = c, j = r; i < j; i++, j--) {
+ const t = stack[i];
stack[i] = stack[j];
stack[j] = t;
}
@@ -939,7 +939,7 @@ class PostScriptEvaluator {
// We can compile most of such programs, and at the same moment, we can
// optimize some expressions using basic math properties. Keeping track of
// min/max values will allow us to avoid extra Math.min/Math.max calls.
-var PostScriptCompiler = (function PostScriptCompilerClosure() {
+const PostScriptCompiler = (function PostScriptCompilerClosure() {
class AstNode {
constructor(type) {
this.type = type;
```
The `this.data` property is, when defined, sent from the worker-thread as a `Uint8Array` and there's thus no reason to re-initialize the TypedArray here.
Note also the `FontFaceObject.createNativeFontFace` method just above, where we simply use `this.data` as-is.
The explanation for this code looking like it does is, as is often the case, for historical reasons. Originally we only supported `@font-face`, before the Font Loading API existed, and back then we also polyfilled TypedArrays (using regular Arrays) which should explain this particular line of code.
Given that the `bytesToString(BaseStream.getBytes(...))` pattern is somewhat common throughout the `src/core/` code, it cannot hurt to add a new `BaseStream`-method which handles that case internally.
- Improve chunking in order to fix some bugs where the spaces aren't here:
* track the last position where a glyph has been drawn;
* when a new glyph (first glyph in a chunk) is added then compare its position with the last saved one and add a space or break:
- there are multiple ways to move the glyphs and to avoid to have to deal with all the different possibilities it's a way easier to just compare positions;
- and so there is now one function (i.e. "compareWithLastPosition") where all the job is done.
- Add some breaks in order to get lines;
- Remove the multiple whites spaces:
* some spaces were filled with several whites spaces and so it makes harder to find some sequences of words using the search tool;
* other pdf readers replace spaces by one white space.
Update src/core/evaluator.js
Co-authored-by: Jonas Jenwald <jonas.jenwald@gmail.com>
Co-authored-by: Jonas Jenwald <jonas.jenwald@gmail.com>
Looking at the `ChunkedStream` implementation, it's basically a "regular" `Stream` but with added functionality in order to deal with fetching/loading of missing data.
Hence, by letting `ChunkedStream` extend `Stream`, we can remove some duplicate methods from the `ChunkedStream` class.
For all of the other `DecodeStream`s we're not passing in a `Dict`-instance manually, but instead get it from the `stream`-parameter. Hence there's no particularly good reason, as far as I can tell, to not do the same thing in `Jbig2Stream`/`JpegStream`/`JpxStream` as well.
The way that `getBaseStreams` is currently handled has bothered me from time to time, especially how we're checking if the method exists before calling it.
By adding a dummy `BaseStream.getBaseStreams` method, and having the call-sites simply check the return value, we can improve some of the relevant code.
Note in particular how the `ObjectLoader._walk` method didn't actually check that the data in question is a Stream instance, and instead only checked the `currentNode` (which could be anything) for the existence of a `getBaseStreams` property.
By having an abstract base-class, it becomes a lot clearer exactly which methods/getters are expected to exist on all Stream instances.
Furthermore, since a number of the methods are *identical* for all Stream implementations, this reduces unnecessary code duplication in the `Stream`, `DecodeStream`, and `ChunkedStream` classes.
For e.g. `gulp mozcentral`, the *built* `pdf.worker.js` files decreases from `1 619 329` to `1 616 115` bytes with this patch-series.
Given that we're using modules, meaning that only explicitly `export`ed things are visible to the outside, it's no longer necessary to wrap all of the code in a closure.
This gets rid of *a lot* of boilerplate that stems from our old way of simulating classes, and it actually reduces the filesize noticeably.
For e.g. `gulp mozcentral`, the *built* `pdf.js` files decreases from `318 404` to `314 722` bytes (~1 percent) with this patch.
This is done automatically with `gulp lint --fix` and the following
manual changes:
```diff
diff --git a/src/core/image.js b/src/core/image.js
index 35c06b8ab..e718b9937 100644
--- a/src/core/image.js
+++ b/src/core/image.js
@@ -97,7 +97,7 @@ class PDFImage {
if (isName(filter)) {
switch (filter.name) {
case "JPXDecode":
- var jpxImage = new JpxImage();
+ const jpxImage = new JpxImage();
jpxImage.parseImageProperties(image.stream);
image.stream.reset();
```
Using `for...of` is a modern and generally much nicer pattern, since it gets rid of unnecessary callback-functions. (In a couple of spots, a "regular" `for` loop had to be used.)
- In case of large string the sandbox initialization failed because of an OOM
* so allocate a new string in the heap
* and free it after use.
- it requires a quickjs update since we need to export some symbols (stringToNewUTF8 and free).
This patch first of all moves all checking/validation into the `appendIfJavaScriptDict` function, to avoid duplicating it in multiple places. Secondly, also removes what's now an outdated/incorrect comment since we have implemented scripting support.
Given that we're (almost) always iterating through the result of the `getAll`-calls, using a `Map` seems nicer overall since it's more suited to iteration compared to a regular Object.
Also, add a couple of `Dict`-checks in existing code touched by this patch, since it really cannot hurt to prevent *potential* errors in a corrupt PDF document.
First of all, while it should be very unlikely that the /ID-entry is an *indirect* object, note how we're using `Dict.get` when parsing it e.g. in `PDFDocument.fingerprint`. Hence we definitely should be consistent here, since if the /ID-entry is an *indirect* object the existing code in `src/core/writer.js` would already fail.
Secondly, to fix the referenced issue, we also need to check that the /ID-entry actually is an Array before attempting to access its contents in `src/core/writer.js`.
*Drive-by change:* In the `xrefInfo` object passed to the `incrementalUpdate` function, re-name the `encrypt` property to `encryptRef` since its data is fetched using `Dict.getRaw` (given the names of the other properties fetched similarly).
For CFF fonts without proper `ToUnicode`/`Encoding` data, utilize the "charset"/"Encoding"-data from the font file to improve text-selection (issue 13260)
Keeps screen readers from pausing on every span so
paragraphs are read more naturally. Note: this only seems
to affect Firefox, Chrome automatically combines the spans.
By not waiting for the /Properties to load, before parsing of the operatorList/textContent starts, there's a very real risk that a `MissingDataException` will be thrown when trying to access the data in the `PartialEvaluator.parseMarkedContentProps` method.
If this ever happens it will thus lead to incomplete and/or outright broken rendering, and with e.g. `disableAutoFetch=true` set the likelihood of this occuring would increase quite a bit.
*Please note:* While I've not yet seen this error in an actual PDF document, it can happen during loading if you're unlucky enough with e.g. the structure of the PDF document and/or the download speed offered by the server.
- Use `PDFManager.ensureDoc`, rather than `PDFManager.ensure`, in a couple of spots in the code. If there exists a short-hand format, we should obviously use it whenever possible.
- Fix a unit-test helper, to account for the previous changes. (Also, converts a function to be `async` instead.)
- Add one more exists-check in `PDFDocument.loadXfaFonts`, which I missed to suggest in PR 13146, to prevent any possible errors if the method is ever called in a situation where it shouldn't be.
Also, print a warning if the actual font-loading fails since that could help future debugging. (Finally, reduce overall indentation in the loop.)
- Slightly unrelated, but make a small tweak of a comment in `src/core/fonts.js` to reduce possible confusion.
- Different fonts can be used in xfa and some of them are embedded in the pdf.
- Load all the fonts in window.document.
Update src/core/document.js
Co-authored-by: Jonas Jenwald <jonas.jenwald@gmail.com>
Update src/core/worker.js
Co-authored-by: Jonas Jenwald <jonas.jenwald@gmail.com>
The size of the `src/core/obj.js` file has increased slowly over the years, and it also contains a fair amount of *distinct* functionality.
In order to improve readability and make it easier to navigate through the code, this patch moves the `XRef` into its own file.
The size of the `src/core/obj.js` file has increased slowly over the years, and it also contains a fair amount of *distinct* functionality.
In order to improve readability and make it easier to navigate through the code, this patch moves `NameTree`/`NumberTree` into its own file.
The size of the `src/core/obj.js` file has increased slowly over the years, and it also contains a fair amount of *distinct* functionality.
In order to improve readability and make it easier to navigate through the code, this patch moves the `FileSpec` into its own file.
The size of the `src/core/obj.js` file has increased slowly over the years, and it also contains a fair amount of *distinct* functionality.
In order to improve readability and make it easier to navigate through the code, this patch moves the `ObjectLoader` into its own file.
`setSelectionRange(0, 0)` added in 44b24fcc29 for #12359, required only by Firefox ([bug](https://bugzilla.mozilla.org/show_bug.cgi?id=860329)), causes issues mozilla#13191, mozilla#12592 in Safari.
`scrollLeft = 0` is a fix that breaks the focus trap in Safari while **keeping Firefox behavior same for #12359**.
Remove the unused "GetIsPureXfa" message handler; and avoid unnecessary parsing when no structTree is available (PR 13069 follow-up, PR 13221 follow-up)
With the current implementation of `PDFDocument.hasJSActions`, in the worker-thread, we're not actually handling not-yet-loaded data correctly. This can thus fail in *two* different ways:
- The `PDFDocument.fieldObjects` getter (and its helper method), while it may *return* a Promise, still fetches all of its data synchronously and it can thus throw a `MissingDataException` during parsing.
- The `Catalog.jsActions` getter, which is completely synchronous, can obviously throw a `MissingDataException` during parsing.
If either of these cases occur currently, the `PDFDocumentProxy.hasJSActions` method in the API can either return a *rejected* Promise (which it never should) or possibly "hang" and never resolve.
*Please note:* While I've not *yet* seen this error in an actual PDF document, it can happen during loading if you're unlucky enough with e.g. the structure of the PDF document and/or the download speed offered by the server.
This patch is thus based on code-inspection *and* on manually throwing a `MissingDataException` on the first access of `Catalog.jsActions` to simulate this situation.
Finally, this patch adds a couple of *API* unit-tests for this (since none existed).
Given that this only an internal helper method, used by the `Catalog.{javaScript, jsActions}` getters, this change simplifies iteration of the returned data.
We can also (slightly) re-factor the code of the `jsActions` getter, and remove an obsolete[1] JSDoc-comment from the `openAction` getter.
---
[1] Not really relevant now that we've got proper scripting support.
Similar to all other data accesses, note e.g. the "GetDocJSActions" handler just above, we need to ensure that a `MissingDataException` isn't propagated to the main-thread if this data is accessed while the PDF document is still loading.
It's obviously (a bit) more efficient to return early in `Page.getStructTree`, rather than trying to first "parse" an *empty* structTree-root.
*Somehow I didn't think of this yesterday, but this feels like a much better solution overall; sorry about the churn here!*
Looking at the API, there's no code which actually sends this message. Most likely it's a left-over from a previous version of PR 13069, since the `isPureXfa` parameter is being included in the "GetDoc" message.
This is first of all consistent with existing API-methods, where we return `null` when the data in question doesn't exist. Secondly, it should also be (slightly) more efficient since there's less dummy-data that we need to transfer between threads.
Finally, this prevents us from adding an empty/unnecessary span to *every* single page even in documents without any structure tree data.
- but don't validate them for now;
- Firefox will display a bar to warn that the signature validation is not supported (see https://bugzilla.mozilla.org/show_bug.cgi?id=854315)
- almost all (all ?) pdf readers display signatures;
- validation is done in edge but for now it's behind a pref.
When a PDF is "marked" we now generate a separate DOM that represents
the structure tree from the PDF. This DOM is inserted into the <canvas>
element and allows screen readers to walk the tree and have more
information about headings, images, links, etc. To link the structure
tree DOM (which is empty) to the text layer aria-owns is used. This
required modifying the text layer creation so that marked items are
now tracked.
Note how we purposely don't expose the `AnnotationStorage`-class directly in the official API (see `src/pdf.js`), since trying to use *multiple* ones simultaneously doesn't really make sense (e.g. in the viewer).
Instead we lazily initialize, and cache, just *one* instance via `PDFDocumentProxy.annotationStorage` which should thus be available internally in the API itself without having to be manually passed to various methods.
To support these changes, the `AnnotationStorage`-instance initialization is moved into the `WorkerTransport`-class to allow both `PDFDocumentProxy` and `PDFPageProxy` to access it.
This patch implements the following simplifications:
- Remove the `annotationStorage`-parameter from `PDFDocumentProxy.saveDocument`, since it's already available internally.
Furthermore, while it's currently possible to call that method without an `AnnotationStorage`-instance, that really does *not* make any sense at all. In this case you're effectively reducing `PDFDocumentProxy.saveDocument` to a "regular" `PDFDocumentProxy.getData` call, but with *a lot* more overhead, which was obviously not the intention of the `PDFDocumentProxy.saveDocument`-method.
- Try to discourage third-party users from calling `PDFDocumentProxy.saveDocument` unconditionally, as a replacement for `PDFDocumentProxy.getData` (note the previous point).
- Replace the `annotationStorage`-parameter, in `PDFPageProxy.render`, with a boolean `includeAnnotationStorage`-parameter which simply indicates if the (internally available) `AnnotationStorage`-instance should be used during rendering (e.g. for printing).
- By removing the need to *manually* provide `annotationStorage`-parameters to various API-methods, using the API should become simpler (e.g. for third-parties) since you no longer need to worry about manually fetching and passing around this data.
The fontName, as defined in the PDF document, cannot be found in *any* of the "name"-tables in the TrueType Collection font. To work-around that, this patch adds a *fallback* code-path to allow using an approximately matching fontName rather than outright failing.
While this method has only been deprecated in one releases now, the `AnnotationStorage`-functionality is new enough that third-party implementations hopefully don't rely heavily on it just yet. (And removing this quickly should help reduce the likelihood that someone starts using it.)
When removing tasks we're currently forced to *indirectly* iterate through the array, which can be avoided by using a Set instead.
Furthermore, we can also (slightly) modernize the code responsible for initializing the `renderTasks`.
Given that what we actually want is only to keep track of the loadedFont-names, rather than storing any actual data, using an object isn't really necessary here. Furthermore, in the current code, we're also using `in` when checking if the data exists, which is generally less efficient than just checking for the value directly.
As mentioned in the JSDoc comment, this should not be used unless you know what you're doing, since it will lead to increased memory usage. However, in some situations (e.g. SVG-rendering), we still want to be able to run general clean-up on both the main/worker-thread while keeping loaded fonts attached to the DOM.[1]
As part of these changes, `WorkerTransport.startCleanup` is converted to an async method and we'll also skip clean-up when destruction has started (since it's redundant).
---
[1] The SVG-rendering mode is obviously not officially supported, since it's both rather incomplete and inherently slower. However with recent changes, whereby we cache repeated images on the document rather than the page level, memory usage can be *a lot* worse than before if we never attempt to release e.g. cached image-data when the viewer is in SVG-rendering mode.
These two properties were *never* intended to be anything but "private", hence it really cannot hurt to actually indicate that they're *not* part of any official API.
Currently the `fontName`-property contains an actual /Name-instance, which is a problem given that its fallback value is an empty string; see ca7f546828/src/core/default_appearance.js (L35)
The reason that this is a problem can be seen in ca7f546828/src/core/primitives.js (L30-L34), since an empty string short-circuits the cache. Essentially, in PDF documents, a /Name-instance cannot be empty and the way that the `DefaultAppearanceEvaluator` does things is unfortunately not entirely correct.
Hence the `fontName`-property is changed to instead contain a string, rather than a /Name-instance, which simplifies the code overall.
*Please note:* I'm tagging this patch with "[api-minor]", since PR 12831 is included in the current pre-release (although we're not using the `fontName`-property in the display-layer).
Fixes#13107
In the issue, some TrueType glyph names have the format `uniXXXX`.
Font's `Encoding` dictionary has the entry `Differences` but no
`BaseEncoding`. `uniXXXX` names are converted to glyph indices
using font's `post` table but currently that is done only when
`BaseEncoding` exists. We must enable the conversion also when only
`Differences` exists.
Currently only URL-strings are officially supported by `getDocument`, however at this point in time I cannot really see any compelling reason to not support `URL`-objects as well.
Most likely the reason that we've don't already support `URL`-objects, in `getDocument`, is that historically `URL` wasn't fully implemented across browsers and our old polyfill wasn't perfect; see https://developer.mozilla.org/en-US/docs/Web/API/URL/URL#browser_compatibility
*Please note:* Because of how the `url` parameter is currently handled, there's actually *some* cases where passing a `URL`-object to `getDocument` already works. That, in my opinion, provides additional motivation for supporting `URL`-objects officially, since it makes the API more consistent.
The following is an attempt to summarize the *current* situation, based on the actual code rather than the JSDocs:
- `getDocument("url string")` works and is documented.[1]
- `getDocument({ url: "url string", })` works and is documented.[1]
- `getDocument(new URL(...))` throws immediately, since no supported parameters are found.
- `getDocument({ url: new URL(...), })` actually works even though it's not documented.[1] Originally, when data was fetched on the worker-thread, this would likely have thrown since `URL` isn't clonable.[2]
- `getDocument({ url: { abc: 123, }, })`, or some similarily meaningless input, will be "accepted" by `getDocument` and then throw a `MissingPDFException` when attempting to fetch the bogus data.
With the changes in this patch, not only is `URL`-objects now officially supported and documented when calling `getDocument`, but we'll also do a much better job at actually validating any URL-data passed to `getDocument` (and instead fail early).
---
[1] In *browsers*, we create a valid URL thus indirectly validating the input. In Node.js environments, on the other hand, no validation is done since obtaining a baseUrl is more difficult (and PDF.js is primarily written for browsers anyway).
[2] https://developer.mozilla.org/en-US/docs/Web/API/Web_Workers_API/Structured_clone_algorithm#supported_types
Given the number of parameters that we now need to parse here, this code is no longer as readable as one would like. Hence this re-factoring, which will improve overall readability and also help with the next patch.
The reasons for making this change are:
- This property is not, nor has it ever been, used anywhere in the PDF.js display-layer.
- Related to the previous point, the format of the `defaultAppearance`-string is such that it'd be difficult to use it as-is in the display-layer anyway.
- It (usually) contains the "raw" appearance-string, from the PDF document, which is neither parsed nor validated and could thus be bogus.
- We now expose a `defaultAppearanceData`-property, which is first of all used in the display-layer and secondly contains actually parsed/validated data.
- In the event that a third-party implementation needs the `defaultAppearance`-string, it could be easily constructed from the recently added `defaultAppearanceData`-property.
All-in-all, I'm thus suggesting that we stop exposing an unused and unnecessary property on all Annotation-instances.
* JS - Handle correctly hierarchy of fields
- it aims to fix#13132;
- annotations can inherit their actions from the parent field;
- there are some fields which act as a container for other fields:
- they can be access through js so need to add them with an empty type (nothing in the spec about that but checked in Acrobat);
- calculation order list (CO) can reference them so need make them through this.getField;
- getArray method must return kids.
- field values are number, string, ... depending of their type but nothing in the spec on how to know what's the type:
- according to the comment for Canonical Format: https://www.adobe.com/content/dam/acom/en/devnet/pdf/pdfs/PDF32000_2008.pdf#page=461
- it seems that this "type" can be guessed from js action Format (when setting a type in Acrobat DC, the only affected thing is this action).
- util.scand with an empty string returns the current date.
Based on this compatibility information, given that IE 11 is now *explicitly* unsupported, we should no longer need to bundle a `URL` polyfill in any builds: https://developer.mozilla.org/en-US/docs/Web/API/URL/URL#browser_compatibility
Note that the caveat listed for older Safari-versions doesn't apply to any code in the PDF.js library, since we never call `new URL(url, undefined)` in the code-base.
Note also that Node.js has a web-compatible `URL` implementation, which according to the "History" section at https://nodejs.org/api/url.html#url_the_whatwg_url_api has been available since Node.js `10.0.0` (according to https://nodejs.org/en/about/releases/ that branch is one month away from being EOL-ed).
The rotation handling that's currently living in `PDFViewerApplication` is *very* old, and pre-dates the introduction of the viewer components by years.
As can be seen in the `BaseViewer.pagesRotation` setter, we're not actually normalizing the rotation as intended and instead rely on the caller to handle that correctly. This is first of all inconsistent, given how other setters are implemented, and secondly it could also lead to the rotation being set to a value outside of the `[0, 360)`-range.
Finally, for improved consistency the rotation handling in `PageViewport` is updated similarly. Please note that this case, it's *not* changing the pre-existing logic.
- implement few positioning properties: position, width, height, anchor;
- implement font element;
- implement fill element (used by font) and its children (linear, radial, ...);
- font property is inherited from ancestor container (see https://www.pdfa.org/wp-content/uploads/2020/07/XFA-3_3.pdf#page=43) so let CSS handles that stuff;
- in order to reduce the number of properties to set, only set non default properties and put the default in CSS;
- set a background to some containers to be able to see them (will be removed in a future commit).
Similar to the existing `annotationsPromise` and `_jsActionsPromise` properties, the new `_xfaPromise` should obviously also be reset, since otherwise you might end up holding onto a lot of data for pages that are no longer active.
(That caching wasn't present in the original version of PR 13069, which is why I didn't spot it until now.)
- add an option to enable XFA rendering if any;
- for now, let the canvas layer: it could be useful to implement XFAF forms (embedded pdf in xml stream for the background and xfa form for the foreground);
- ui elements in template DOM are pretty close to their html counterpart so we generate a fake html DOM from template one:
- it makes easier to translate template properties to html ones;
- it makes faster the creation of the html element in the main thread.
While there is nothing *outright* wrong with the existing implementation, it can however lead to increased memory usage in one particular case (that I completely overlooked when implementing this):
For "data:"-URLs, which by definition contains the entire PDF document and can thus be arbitrarily large, we obviously want to avoid sending, storing, and/or logging the "raw" docBaseUrl in that case.
To address this, this patch makes the following changes:
- Ignore any non-string in the `docBaseUrl` option passed to `getDocument`, since those are unsupported anyway, already on the main-thread.
- Ignore "data:"-URLs in the `docBaseUrl` option passed to `getDocument`, to avoid having to send what could potentially be a *very* long string to the worker-thread.
- Parse the `docBaseUrl` option *directly* in the `BasePdfManager`-constructors, on the worker-thread, to avoid having to store the "raw" docBaseUrl in the first place.
It seems reasonable to place this alongside the *similar* `getFilenameFromUrl` helper function. This way, with the changes in the next patch, we also avoid having to expose the `isDataScheme` function in the API itself and we instead expose `getPdfFilenameFromUrl` in the API (which feels overall more appropriate).
This extends PR 13033 slightly, with a heuristic to support corrupt PDF documents where the `LineAnnotation`s have an empty /Rect-entry. Please note that while I have no idea if this is "correct", this patch at least makes us output the same /BBox as re-saving in Adobe Reader does.
This is mostly done using `gulp lint --fix` with a few manual changes in
the following diff:
```diff
diff --git a/src/core/pattern.js b/src/core/pattern.js
index 365491ed3..eedd8b686 100644
--- a/src/core/pattern.js
+++ b/src/core/pattern.js
@@ -105,7 +105,7 @@ const Pattern = (function PatternClosure() {
return Pattern;
})();
-var Shadings = {};
+const Shadings = {};
// A small number to offset the first/last color stops so we can insert ones to
// support extend. Number.MIN_VALUE is too small and breaks the extend.
@@ -597,16 +597,15 @@ Shadings.Mesh = (function MeshClosure() {
if (!(0 <= f && f <= 3)) {
throw new FormatError("Unknown type6 flag");
}
- var i, ii;
const pi = coords.length;
- for (i = 0, ii = f !== 0 ? 8 : 12; i < ii; i++) {
+ for (let i = 0, ii = f !== 0 ? 8 : 12; i < ii; i++) {
coords.push(reader.readCoordinate());
}
const ci = colors.length;
- for (i = 0, ii = f !== 0 ? 2 : 4; i < ii; i++) {
+ for (let i = 0, ii = f !== 0 ? 2 : 4; i < ii; i++) {
colors.push(reader.readComponents());
}
- var tmp1, tmp2, tmp3, tmp4;
+ let tmp1, tmp2, tmp3, tmp4;
switch (f) {
// prettier-ignore
case 0:
@@ -729,16 +728,15 @@ Shadings.Mesh = (function MeshClosure() {
if (!(0 <= f && f <= 3)) {
throw new FormatError("Unknown type7 flag");
}
- var i, ii;
const pi = coords.length;
- for (i = 0, ii = f !== 0 ? 12 : 16; i < ii; i++) {
+ for (let i = 0, ii = f !== 0 ? 12 : 16; i < ii; i++) {
coords.push(reader.readCoordinate());
}
const ci = colors.length;
- for (i = 0, ii = f !== 0 ? 2 : 4; i < ii; i++) {
+ for (let i = 0, ii = f !== 0 ? 2 : 4; i < ii; i++) {
colors.push(reader.readComponents());
}
- var tmp1, tmp2, tmp3, tmp4;
+ let tmp1, tmp2, tmp3, tmp4;
switch (f) {
// prettier-ignore
case 0:
@@ -897,7 +895,7 @@ Shadings.Mesh = (function MeshClosure() {
decodeType4Shading(this, reader);
break;
case ShadingType.LATTICE_FORM_MESH:
- var verticesPerRow = dict.get("VerticesPerRow") | 0;
+ const verticesPerRow = dict.get("VerticesPerRow") | 0;
if (verticesPerRow < 2) {
throw new FormatError("Invalid VerticesPerRow");
}
```
A significant portion of the code-base has now been converted to use `let`/`const`, rather than `var`, hence it should be possible to simply enable the ESLint `no-var` rule globally.
This way we can ensure that new code won't accidentally use `var`, and it also removes the need to manually enable the rule in various folders.
Obviously it makes sense to continue the efforts to replace `var`, but that should probably happen on a file and/or folder basis.
Please note that this patch excludes the following code:
- The `extensions/` folder, since that seemed easiest for now (and I don't know exactly what the support situation is for the Chromium-extension).
- The entire `external/` folder is ignored, since most of it's currently excluded from linting.
For the code that isn't imported from elsewhere (and should be ignored), we should probably (at some point) bring the code up to the same linting/formatting standard as the rest of the code-base.
- Various files in the `test/` folder are ignored, as necessary, since the way that a lot of this code is loaded will require some care (or perhaps larger re-factoring) when removing `var` usage.
While the JSDocs have never advertised `getDocument` as supporting Node.js `Buffer`s, that apparently doesn't stop users from passing such data structures to `getDocument`.
In theory the existing `instanceof Uint8Array` check ought to have caught Node.js `Buffer`s, however for reasons that I don't even pretend to understand that check actually passes. Hence this patch which, *only* in Node.js environments, will special-case `Buffer`s to hopefully provide a slightly better out-of-the-box behaviour in Node.js environments[1].
---
[1] Although I'm not sure that we necessarily want to advertise this in the JSDocs, given the specialized use-case.
Replace the `objectFromEntries` helper function with an `objectFromMap` one instead, and simplify the data lookup in the AnnotationStorage.getValue method
Note that the majority of these changes were done automatically, by using `gulp lint --fix`, and the manual changes were limited to the following diff:
```diff
diff --git a/src/core/cff_parser.js b/src/core/cff_parser.js
index d684c200e..2e2b811e4 100644
--- a/src/core/cff_parser.js
+++ b/src/core/cff_parser.js
@@ -555,7 +555,7 @@ const CFFParser = (function CFFParserClosure() {
stackSize %= 2;
validationCommand = CharstringValidationData[value];
} else if (value === 10 || value === 29) {
- var subrsIndex;
+ let subrsIndex;
if (value === 10) {
subrsIndex = localSubrIndex;
} else {
@@ -886,15 +886,15 @@ const CFFParser = (function CFFParserClosure() {
format = bytes[pos++];
switch (format & 0x7f) {
case 0:
- var glyphsCount = bytes[pos++];
+ const glyphsCount = bytes[pos++];
for (i = 1; i <= glyphsCount; i++) {
encoding[bytes[pos++]] = i;
}
break;
case 1:
- var rangesCount = bytes[pos++];
- var gid = 1;
+ const rangesCount = bytes[pos++];
+ let gid = 1;
for (i = 0; i < rangesCount; i++) {
const start = bytes[pos++];
const left = bytes[pos++];
@@ -938,7 +938,7 @@ const CFFParser = (function CFFParserClosure() {
}
break;
case 3:
- var rangesCount = (bytes[pos++] << 8) | bytes[pos++];
+ const rangesCount = (bytes[pos++] << 8) | bytes[pos++];
for (i = 0; i < rangesCount; ++i) {
let first = (bytes[pos++] << 8) | bytes[pos++];
if (i === 0 && first !== 0) {
@@ -1173,7 +1173,7 @@ class CFFDict {
}
}
-var CFFTopDict = (function CFFTopDictClosure() {
+const CFFTopDict = (function CFFTopDictClosure() {
const layout = [
[[12, 30], "ROS", ["sid", "sid", "num"], null],
[[12, 20], "SyntheticBase", "num", null],
@@ -1229,7 +1229,7 @@ var CFFTopDict = (function CFFTopDictClosure() {
return CFFTopDict;
})();
-var CFFPrivateDict = (function CFFPrivateDictClosure() {
+const CFFPrivateDict = (function CFFPrivateDictClosure() {
const layout = [
[6, "BlueValues", "delta", null],
[7, "OtherBlues", "delta", null],
@@ -1265,11 +1265,12 @@ var CFFPrivateDict = (function CFFPrivateDictClosure() {
return CFFPrivateDict;
})();
-var CFFCharsetPredefinedTypes = {
+const CFFCharsetPredefinedTypes = {
ISO_ADOBE: 0,
EXPERT: 1,
EXPERT_SUBSET: 2,
};
+
class CFFCharset {
constructor(predefined, format, charset, raw) {
this.predefined = predefined;
@@ -1695,7 +1696,7 @@ class CFFCompiler {
// For offsets we just insert a 32bit integer so we don't have to
// deal with figuring out the length of the offset when it gets
// replaced later on by the compiler.
- var name = dict.keyToNameMap[key];
+ const name = dict.keyToNameMap[key];
// Some offsets have the offset and the length, so just record the
// position of the first one.
if (!offsetTracker.isTracking(name)) {
```
Rather than first checking if data exists before fetching it from storage, we can simply do the lookup directly and then check its value.
Note that this follows the same pattern as utilized in the `AnnotationStorage.setValue` method.
Given that it's only used with `Map`s, and that it's currently implemented in such a way that we (indirectly) must iterate through the data *twice*, some simplification cannot hurt here.
Note that the only reason that we're not using `Object.fromEntries(...)` directly, at each call-site, is that that one won't guarantee that a `null` prototype is being used.
Now that we have scripting support, warning about e.g. JavaScript actions doesn't seem necessary anymore. Especially considering that scripting-related actions are/will not be parsed by the `Catalog.parseDestDictionary` method anyway, since it's intended for handling "simple" actions.
All of this code predates the existence of native JS classes, however we can now clean this up a bit. This patch thus let us remove some variable "shadowing" from the code.
This helper function is first of all only called *twice*, and secondly it also leads to unnecessary intermediate allocations given how the `TypedArray`s are handled.
Hence we can simply inline this small function, and thus directly allocate the combined `TypedArray` instead.
The `compareByteArrays` is first of all duplicated in multiple closures in the `src/core/crypto.js` file. Secondly, despite its name, it's also functionally equivalent to the now existing `isArrayEqual` helper function.
The `isArrayEqual` helper function is changed to use a standard `for`-loop, rather than `Array.prototype.every`, since that ought to be slightly more efficient given that we're now using it with (potentially) larger data.
All of this code predates the existence of native JS classes, however we can now clean this up a bit. This patch thus let us remove some variable "shadowing" from the code.
Note that this particular helper function is, with the exception of the `GENERIC` default viewer and the (unsupported) SVG-backend, mostly unused at this point in time. Hence we should be able to clean-up this helper function slightly.
Also, fixes a small inconsistency in the `SVGGraphics` initialization in the viewer, by passing in the `disableCreateObjectURL` compatibility-option. Given that the SVG-backend isn't officially supported/recommended this shouldn't have been an issue, but given that I spotted this it can't hurt to fix it.
Note how the `PDFAttachmentViewer` handles PDF file attachments specially, by opening them in a new window/tab, rather than forcing them to be downloaded. This is done to improve the overall UX, since browsers in general are able to handle PDF files internally.
However, for file *annotations* we're currently not attempting to do the same thing and are instead just downloading them directly. In order to unify the behaviour, without having to duplicate a lot of code, the opening of PDF file attachments is thus moved into a new `DownloadManager.openOrDownloadData` method.
- strokeColor corresponds to borderColor;
- support fillColor and textColor;
- support colors on the different annotations;
- fix typo in aforms (+test).
This is similar to the other methods, and the only reason for this not having been done originally is that the `cancel` functionality is a later addition.
Rather than converting the `AnnotationStorage`-data to an Object, before sending it to the worker-thread, we should be able to simply send the internal `Map` directly.
The "structured clone algorithm" doesn't have a problem with `Map`s, however the `LoopbackPort` used when workers are *disabled* (e.g. in Node.js environments) didn't use to support them. With PR 12997 having lifted that restriction, we should now be able to simply send the `AnnotationStorage`-data as-is rather than having to iterate through it to first create an Object.
*Please note:* The changes in `src/core/annotation.js` could have been a lot more compact if we were able to use optional chaining in the `src/core` folder. Unfortunately that's still not possible, since SystemJS is being used in the development viewer (i.g. `gulp server`) and fixing that is *still* blocked by [bug 1247687](https://bugzilla.mozilla.org/show_bug.cgi?id=1247687).
*Please note:* The `defer` parameter has been enabled by default ever since PR 9777 (in 2018), which first shipped in PDF.js release `2.0.943`.
With workers *disabled*, e.g. in Node.js environments, this has been used ever since without any problems reported[1].
The impetus for this change was that I happened to notice that *if* the `LoopbackPort` was used with synchronous event dispatching, we'd simply send that data as-is to the listeners. This created an inconsistency in the data returned from the `pdf.worker.js` file, since `postMessage` used with *actual* workers (or the `LoopbackPort` with `defer = true`) will ignore/throw when encountering unclonable data.
Originally my intention was simply to just call `cloneValue` regardless of the event dispatching used in `LoopbackPort`, however looking at the use-cases (or lack thereof) of the `LoopbackPort` it seemed reasonable to simply remove the `defer` parameter instead.
This patch is tagged "[api-minor]" since the `LoopbackPort` is still exposed in the API, although I really hope that no third-party is using this (since disabling workers leads to bad performance).
Finally, this patch changes a `forEach` loop to `for...of` and makes uses of optional changing in existing code.
---
[1] As evident by the `npm test` command run by Github Actions, and previously by Travis.
With the previous patch this function is now *only* accessed on the worker-thread, hence it's no longer necessary to include it in the *built* `pdf.js` file.
With the previous patch this functionality is now *only* accessed on the worker-thread, hence it's no longer necessary to include it in the *built* `pdf.js` file.
The only reason, as far as I can tell, for parsing the Metadata on the main-thread is how it was originally implemented. When Metadata support was first implemented, it utilized the [`DOMParser`](https://developer.mozilla.org/en-US/docs/Web/API/DOMParser) which isn't available in workers.
Today, with the custom XML-parser being used, that's no longer an issue and it seems reasonable to move the Metadata parsing to the worker-thread[1], since that's where all parsing should happen (for performance reasons).
Based on these changes, we'll be able to reduce the now unnecessary duplication of the XML-parser (and related code) in both of the *built* `pdf.js`/`pdf.worker.js` files.
Finally, this patch changes the `_repair` method to use "Array + join" rather than string concatenation.
---
[1] This needed the previous patch, to enable sending of `Map`s between threads with workers disabled.
I happened to look at this code, and I can't for the life of me figure out why I didn't just implement it like this patch in the first place (since the current format feels overly verbose).
The following checks are all unneeded, and could easily cause confusion when reading the code. (All of them are my fault as well, since I've sometimes added those checks without really thinking about the surrounding code.)
- In `PartialEvaluator.hasBlendModes` there cannot be any `MissingDataException`s thrown, given that the `Page.getOperatorList` method waits for all the necessary /Resources to load first. Furthermore, note also that if an error is thrown from `PartialEvaluator.hasBlendModes` then it'd completely break rendering of that page, since any errors thrown from `Page.getOperatorList` are simply sent to the main-thread.
- In `PartialEvaluator.handleColorN` there cannot be any `MissingDataException`s thrown, given that again the `Page.getOperatorList` method waits for all the necessary /Resources to load before operatorList parsing starts.
- In `XRef.readXRef` there cannot be any `MissingDataException`s thrown, given that we're *explicitly* requesting (and waiting for) the entire document in `pdfManagerReady` (in `src/core/worker.js`) before re-parsing of a corrupt document starts.
* don't set a value in annotationStorage by default:
- having an undefined when the annotation is rendered for saving/printing means nothing has changed so use normal appearance
- aims to fix https://bugzilla.mozilla.org/show_bug.cgi?id=1681687
* change the way to compute font size when this one is null in DA:
- make fontSize proportional to line height
- in multiline case, take into account the number of lines for text entered to adapt the font size
- use ascent of the fallback font instead of the one from pdf to position spans
- use TextMetrics.fontBoundingBoxAscent if available or
- use a basic heuristic to guess ascent in drawing char on a canvas
- compute ascent as a ratio of font height
The *third* page of the referenced PDF document currently fails to render completely, since one of its font files fail to load.
Since that error isn't handled, a large part of the text is thus missing which looks quite bad. By "replacing" the font data with an *empty* stream, we'll thus be able to fallback to rendering the text with a standard font (instead of using `ErrorFont`). While there's obviously no guarantee that things will look perfect, actually rendering the text at all should be an improvement in general.
Also, print a warning in `PartialEvaluator.loadFont` when the `PartialEvaluator.translateFont` method rejects, since that'd have helped debug/fix the issue faster.
*As far as I can tell, this has been broken ever since PR 3289 (back in 2013) without anyone noticing.*
For any non-`MissingDataException` errors encountered in `ObjectLoader._walk`, we're simply throwing immediately which thus has the potential to *completely* break rendering of an entire page.
In practice this is obviously only an issue for PDF documents which are in one way or another corrupt, since that's the only way that `XRef.fetch` will throw non-`MissingDataException` errors. To make matters worse these errors are *intermittent*, since they can only occur if the document is still loading when the `ObjectLoader`-code runs (note the early return in `ObjectLoader.load`).
Please note that we cannot simply catch the error and let "normal" parsing continue in `ObjectLoader._walk`, since that could lead to errors elsewhere given that resources "below" the current one (in the graph) might not be checked as intended then.
All-in-all, the only way to make absolutely sure that we won't cause *unexpected* `MissingDataException`s somewhere else in the code-base is to fallback to fetching the *entire* document in this edge-case.
- in order to evaluate SOM expressions nodes and their attributes must be checked in the same order as in the xml;
- add an object XFAObjectArray with a parameter max to handle multiple children with the same name.
- the parser is base on a class extending XMLParserBase
- it handle xml namespaces:
* each namespace is assocated with a builder
* builder builds nodes belonging to the namespace
* when a node is inserted in the parent namespace compatibility is checked (if required)
- to avoid name collision between xml names and object properties, use Symbol.
Given that `FontFaceObject` is not exposed in the public API, but only accessed internally, there's no need to assume that a `FontFaceObject`-instance is ever initialized without `onUnsupportedFeature` being provided. This is also consistent with the `BaseFontLoader` implementation.
Given that we'll only cache `/XObject`s of the `Image`-type globally, we can utilize that in `PartialEvaluator.getTextContent` as well. This way, in cases such as e.g. issue 12098, we can avoid having to fetch/parse `/XObject`s that we already know to be `Image`s. This is helpful, since `Stream`s are not cached on the `XRef` instance (given their potential size) and the lookup can thus be somewhat expensive in general.
Also, skip a redundant `RefSetCache.has` check in the `GlobalImageCache.getData` method.
The widths property should be an object to match what metrics returns.
In ZapfDingbats.pdf I was getting a data clone error with pdfBug enabled.
In buildCharCodeToWidth() there was an encoding with the name "at" which
is also the name of a method on an array. buildCharCodeToWidth assumes an
object is passed in, so when it checked for the "at" property, it found the
method and copied it over.
This only seemed to affect Firefox.
When implementing the `GlobalImageCache` functionality I was mostly worried about the effect of *very large* images, hence the maximum number of cached images were purposely kept quite low[1].
However, there's one fairly obvious problem with that approach: In documents with hundreds, or even thousands, of *small* images the `GlobalImageCache` as implemented becomes essentially pointless.
Hence this patch, where the `GlobalImageCache`-implementation is changed in the following ways:
- We're still guaranteed to be able to cache a *minimum* number of images, set to `10` (similar as before).
- If the *total* size of all the cached image data is below a threshold[2], we're allowed to cache additional images.
This patch thus *improve*, but doesn't completely fix, issue 12098. Note that that document is created by a *very poor* PDF generator, since every single page contains the *entire* document (with all of its /Resources) and to create the individual pages clipping is used.[3]
---
[1] Currently set to `10` images; imagine what would happen to overall memory usage if we encountered e.g. 50 images each 10 MB in size.
[2] This value was chosen, somewhat randomly, to be `40` megabytes; basically five times the [maximum individual image size per page](6249ef517d/src/display/api.js (L2483-L2484)).
[3] This surely has to be some kind of record w.r.t. how badly PDF generators can mess things up...
Note how, in the `if (this.stateManager.stateStack.length !== 0) {` branch, we're attempting to access the not yet defined variable[1] `args`. If this code-path is ever hit, an Error will be thrown and parsing will thus be aborted immediately (likely leading to e.g. rendering bugs).
Note that I found this purely by accident, since I happened to glance at the LGTM report. However, I've since found that the error is also present during the unit-test[2] and with this patch we're actually testing the *intended* thing here.
As part of fixing this, and to avoid re-introducing a similar bug in the future, we'll now instead always reset `args.length` *before* attempting to read the next operator.
Also, we can use the existing `EvaluatorPreprocessor.savedStatesDepth` getter to simplify the save/restore detection a tiny bit.
---
[1] The ESLint rule `no-use-before-define` would have helped catch this problem, but unfortunately we cannot enable that without quite a bit of refactoring all over the code-base.
[2] The unit-test was updated such that it would fail in the `master`-branch.
With the changes in PR 12831, it's no longer necessary to keep track of the `fontName`-string separately since it's available through the `_defaultAppearanceData`-property as well.
As can be seen in 2cba290361/src/core/evaluator.js (L986) the `gStateObj` (which is actually an Array despite its name), is wrapped in Array when it's inserted into the OperatorList. Hence we obviously need to take this into account when accessing it in `TranslatedFont._removeType3ColorOperators`; this mistake happened because we don't have any test-cases for this particular code-path as far as I know.
By changing this a `shadow`ed getter, we can simply access it directly and not worry about it being initialized. I have no idea why I didn't just implement it this way in the first place.
* Add a parser to get font data from the default appearance
- pdfium & poppler use a special parser too to get these info.
* Update src/core/default_appearance.js
Co-authored-by: Jonas Jenwald <jonas.jenwald@gmail.com>
Co-authored-by: Jonas Jenwald <jonas.jenwald@gmail.com>
- aims to fix issue #12868: apply zoom factor to linewidth after setting it to 1.
- only apply 1px-width when required
- the sign of getSinglePixelWidth is used to know if 1px-width is required
- Remove a *duplicated* reference test, see "issue12810", from the manifest.
- Improve the spelling in a couple of comments in `src/core/canvas.js`, most notable of the word "parallelogram".
- Update a comment, also in `src/core/canvas.js`, to actually agree with the value used to reduce confusion when reading the code.
While PR 12725 fixed bug 1671312 as reported, i.e. the "In the upper right corner "Purposes' has bad kerning."-part, it however broke other parts of the text rendering.
Note in particular the tables, e.g. on page 2 and beyond, where the glyphs are now rendered too close together. The reason for this is that the fonts in question are non-embedded ArialNarrow, which we just replace with Helvetica which obviously is not narrow. Given that the font replacement isn't a perfect fit for non-embedded ArialNarrow, we still need to re-measure the glyph widths in this case.
Note first of all how the `PDFDocumentProxy.getJSActions` method in the API caches the result, which makes repeated lookups cheap enough to not really be an issue.
Secondly, with the previous patch, we're now only dispatching "pageopen"/"pageclose"-events when there's actually a sandbox that listens for them.
All-in-all, with these changes we can thus simplify the default-viewer "pageopen"-event handler a fair bit.
* add a comment to explain how minimal linewidth is computed.
* when context.linewidth < 1 after transform, firefox and chrome
don't render in the same way (issue #12810).
* set lineWidth to 1 after transform and before stroking
- aims fix issue #12295
- a pixel can be transformed into a rectangle with both heights < 1.
A single rescale leads to a rectangle with dim equals to 1 and
the other to something greater than 1.
* change the way to render rectangle with null dimensions:
- right now we rely on the lineWidth set before "re" but
it can be set after "re" and before "S" and in this case the rendering
will be wrong.
- render such rectangles as a single line.
There's built-in ESLint rule, see `sort-imports`, to ensure that all `import`-statements are sorted alphabetically, since that often helps with readability.
Unfortunately there's no corresponding rule to sort `export`-statements alphabetically, however there's an ESLint plugin which does this; please see https://www.npmjs.com/package/eslint-plugin-sort-exports
The only downside here is that it's not automatically fixable, but the re-ordering is a one-time "cost" and the plugin will help maintain a *consistent* ordering of `export`-statements in the future.
*Note:* To reduce the possibility of introducing any errors here, the re-ordering was done by simply selecting the relevant lines and then using the built-in sort-functionality of my editor.
Given that the PDF document in the issue contains the same very large JPEG image *three* times, this patch includes a test-case where only the first page has been extracted from it.
Given that the API will now, after PR 12039, automatically pick the correct factories to use depending on the environment (browser vs. Node.js), we can utilize that in the unit-tests as well. This way we don't have to manually repeat the same initialization code in *multiple* unit-tests.
*Note:* The *official* PDF.js API is defined in `src/pdf.js`, hence the new exports in `src/display/api.js` will not affect that.
Also, updates the unit-test `FileReaderFactory` helpers similarily.
*Drive-by change:* Fix the `CMapReaderFactory` usage in the annotation unit-tests, since the cache should only contain raw data and not a Promise. While this obviously works as-is, having unit-tests that "abuse" the intended data format can easily lead to unnecessary failures if changes are made to the relevant `src/core/` code.
