mirrors/pdf.js
1
0
Fork 0
mirror of https://github.com/mozilla/pdf.js.git synced 2025-04-19 06:38:07 +02:00
Code Activity

Allow searching for number-number on two lines

Browse source 
When a dash separates two digits, it's very likely to not be a hyphen
inserted to split a word into two lines (e.g. "par\n-ser"), but rather
either a minus sign, a range, or a date. For example, in the tracemonkey
PDF there is `2008-02` (a date) split across two lines.

Preserving the dash, similarly to how we do for compound words, allows
searches for "2008-02" to find a match.
This commit is contained in:
Nicolò Ribaudo 2025-01-14 17:54:37 +01:00
parent 016de74229
commit 8358ab63b3
No known key found for this signature in database
GPG key ID: AAFDA9101C58F338
2 changed files with 78 additions and 30 deletions
Download patch file Download diff file Expand all files Collapse all files

34
test/unit/pdf_find_controller_spec.js
View file

@ -1104,6 +1104,40 @@ describe("pdf_find_controller", function () {
});
});
it("performs a search with a dash between two digits", async () => {
const { eventBus, pdfFindController } = await initPdfFindController();
await testSearch({
eventBus,
pdfFindController,
state: {
query: "2008-02",
},
matchesPerPage: [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1],
selectedMatch: {
pageIndex: 13,
matchIndex: 0,
},
pageMatches: [[], [], [], [], [], [], [], [], [], [], [], [], [], [314]],
pageMatchesLength: [
[],
[],
[],
[],
[],
[],
[],
[],
[],
[],
[],
[],
[],
[7],
],
});
});
describe("custom matcher", () => {
it("calls to the matcher with the right arguments", async () => {
const QUERY = "Foo bar";

74
web/pdf_find_controller.js
View file

@ -117,10 +117,12 @@ function normalize(text) {
}
}
const hasSyllables = syllablePositions.length > 0;
let normalizationRegex;
if (syllablePositions.length === 0 && noSyllablesRegExp) {
if (!hasSyllables && noSyllablesRegExp) {
normalizationRegex = noSyllablesRegExp;
} else if (syllablePositions.length > 0 && withSyllablesRegExp) {
} else if (hasSyllables && withSyllablesRegExp) {
normalizationRegex = withSyllablesRegExp;
} else {
// Compile the regular expression for text normalization once.
@ -131,22 +133,33 @@ function normalize(text) {
// 30A0-30FF: Katakana
const CJK = "(?:\\p{Ideographic}|[\u3040-\u30FF])";
const HKDiacritics = "(?:\u3099|\u309A)";
const CompoundWord = "\\p{Ll}-\\n\\p{Lu}";
const regexp = `([${replace}])|([${toNormalizeWithNFKC}])|(${HKDiacritics}\\n)|(\\p{M}+(?:-\\n)?)|(${CompoundWord})|(\\S-\\n)|(${CJK}\\n)|(\\n)`;
const BrokenWord = `\\p{Ll}-\\n(?=\\p{Ll})|\\p{Lu}-\\n(?=\\p{L})`;
if (syllablePositions.length === 0) {
// Most of the syllables belong to Hangul so there are no need
// to search for them in a non-Hangul document.
// We use the \0 in order to have the same number of groups.
normalizationRegex = noSyllablesRegExp = new RegExp(
regexp + "|(\\u0000)",
"gum"
);
const regexps = [
/* p1 */ `[${replace}]`,
/* p2 */ `[${toNormalizeWithNFKC}]`,
/* p3 */ `${HKDiacritics}\\n`,
/* p4 */ "\\p{M}+(?:-\\n)?",
/* p5 */ `${BrokenWord}`,
/* p6 */ "\\S-\\n",
/* p7 */ `${CJK}\\n`,
/* p8 */ "\\n",
/* p9 */ hasSyllables
? FIRST_CHAR_SYLLABLES_REG_EXP
: // Most of the syllables belong to Hangul so there are no need
// to search for them in a non-Hangul document.
// We use the \0 in order to have the same number of groups.
"\\u0000",
];
normalizationRegex = new RegExp(
regexps.map(r => `(${r})`).join("|"),
"gum"
);
if (hasSyllables) {
withSyllablesRegExp = normalizationRegex;
} else {
normalizationRegex = withSyllablesRegExp = new RegExp(
regexp + `|(${FIRST_CHAR_SYLLABLES_REG_EXP})`,
"gum"
);
noSyllablesRegExp = normalizationRegex;
}
}
@ -281,26 +294,27 @@ function normalize(text) {
}
if (p5) {
// Compound word with a line break after the hyphen.
// Since the \n isn't in the original text, o = 3 and n = 3.
shiftOrigin += 1;
eol += 1;
return p5.replace("\n", "");
}
if (p6) {
// "X-\n" is removed because an hyphen at the end of a line
// with not a space before is likely here to mark a break
// in a word.
// In "X-\ny", "-\n" is removed because an hyphen at the end of a line
// between two letters is likely here to mark a break in a word.
// If X is encoded with UTF-32 then it can have a length greater than 1.
// The \n isn't in the original text so here y = i, n = X.len - 2 and
// o = X.len - 1.
const len = p6.length - 2;
const len = p5.length - 2;
positions.push(i - shift + len, 1 + shift);
shift += 1;
shiftOrigin += 1;
eol += 1;
return p6.slice(0, -2);
return p5.slice(0, -2);
}
if (p6) {
// A - following a non-space character that is not detected as the
// hyphen breaking a word in two lines needs to be preserved. It could
// be, for example, in a compound word or in a date.
// Only remove the newline.
shiftOrigin += 1;
eol += 1;
return p6.slice(0, -1);
}
if (p7) {
@ -324,7 +338,7 @@ function normalize(text) {
return " ";
}
// p8
// p9
if (i + eol === syllablePositions[syllableIndex]?.[1]) {
// A syllable (1 char) is replaced with several chars (n) so
// newCharsLen = n - 1.