@angri450/zvec-grep 0.2.2

package/dist/engine/extraction/markdown/extractor.js

@@ -0,0 +1,362 @@
+import { EngineError } from "../../errors.js";
+import { makeEntityId } from "../ids.js";
+import { validateSourceFile } from "../source.js";
+import { extractPlainTextFragments } from "../text/extractor.js";
+import { chunkOptionsForMetadata, fitTextToChars } from "../vector-content.js";
+const DEFAULT_MARKDOWN_CHUNK_CHARS = 3600;
+const DEFAULT_MARKDOWN_CHUNK_OVERLAP_CHARS = 540;
+export class MarkdownExtractor {
+    async extract(source, options = {}) {
+        if (source.kind !== "text" || source.file.format !== "markdown") {
+            return [];
+        }
+        validateSourceFile(source);
+        const chunkOptions = resolveMarkdownChunkOptions(options);
+        const lines = source.text.split("\n");
+        const headings = scanHeadings(lines);
+        if (headings.length === 0) {
+            return this.fallback(source, chunkOptions);
+        }
+        const lineOffsets = computeLineOffsets(lines);
+        const fenceLines = computeFenceLines(lines);
+        const sections = buildSections(headings, lines);
+        const fragments = [];
+        for (const section of sections) {
+            const metadata = markdownMetadata(section);
+            const contentChunkOptions = chunkOptionsForMetadata(chunkOptions, metadata);
+            const windows = splitMarkdownSection({
+                lines,
+                lineOffsets,
+                fenceLines,
+                section,
+                maxChars: contentChunkOptions.maxChunkChars,
+                overlapChars: contentChunkOptions.chunkOverlapChars,
+            });
+            if (windows.length > 1) {
+                const id = makeEntityId(source.file.id, fragments.length);
+                fragments.push({
+                    id,
+                    group: id,
+                    fileId: source.file.id,
+                    range: linesToWindow(lines, lineOffsets, section.startIndex, section.endIndex).range,
+                    content: {
+                        kind: "text",
+                        text: fitTextToChars(markdownOutline(metadata), contentChunkOptions.maxChunkChars),
+                    },
+                    metadata,
+                });
+                for (const window of windows) {
+                    fragments.push(markdownWindowToFragment(source, metadata, window, fragments.length, id));
+                }
+                continue;
+            }
+            for (const window of windows) {
+                fragments.push({
+                    id: makeEntityId(source.file.id, fragments.length),
+                    fileId: source.file.id,
+                    range: window.range,
+                    content: {
+                        kind: "text",
+                        text: window.text,
+                    },
+                    metadata,
+                });
+            }
+        }
+        return fragments.length > 0
+            ? fragments
+            : this.fallback(source, chunkOptions);
+    }
+    fallback(source, options) {
+        return extractPlainTextFragments(source, options.maxChunkChars, options.chunkOverlapChars);
+    }
+}
+function resolveMarkdownChunkOptions(options) {
+    const maxChunkChars = options.maxChunkChars ?? DEFAULT_MARKDOWN_CHUNK_CHARS;
+    const chunkOverlapChars = options.chunkOverlapChars ?? DEFAULT_MARKDOWN_CHUNK_OVERLAP_CHARS;
+    if (!Number.isInteger(maxChunkChars) || maxChunkChars <= 0) {
+        throw new EngineError("Markdown extractor requires a positive integer chunk size", {
+            code: "ZVEC_GREP.ENGINE.EXTRACTORS.MARKDOWN_INVALID_CHUNK_SIZE",
+            context: `maxChunkChars=${maxChunkChars}`,
+        });
+    }
+    if (!Number.isInteger(chunkOverlapChars) ||
+        chunkOverlapChars < 0 ||
+        chunkOverlapChars >= maxChunkChars) {
+        throw new EngineError("Markdown extractor requires overlap to be smaller than chunk size", {
+            code: "ZVEC_GREP.ENGINE.EXTRACTORS.MARKDOWN_INVALID_CHUNK_OVERLAP",
+            context: `maxChunkChars=${maxChunkChars} chunkOverlapChars=${chunkOverlapChars}`,
+        });
+    }
+    return { maxChunkChars, chunkOverlapChars };
+}
+function markdownWindowToFragment(source, metadata, window, index, group) {
+    return {
+        id: makeEntityId(source.file.id, index),
+        group,
+        fileId: source.file.id,
+        range: window.range,
+        content: {
+            kind: "text",
+            text: window.text,
+        },
+        metadata,
+    };
+}
+function markdownOutline(metadata) {
+    return metadata.heading ?? "markdown section";
+}
+function scanHeadings(lines) {
+    const headings = [];
+    let fence = null;
+    for (let index = 0; index < lines.length; index++) {
+        const line = lines[index];
+        const trimmed = line.trimStart();
+        if (fence) {
+            if (trimmed.startsWith(fence)) {
+                fence = null;
+            }
+            continue;
+        }
+        if (trimmed.startsWith("```")) {
+            fence = "```";
+            continue;
+        }
+        if (trimmed.startsWith("~~~")) {
+            fence = "~~~";
+            continue;
+        }
+        const atx = line.match(/^(#{1,6})\s+(.+?)\s*#*\s*$/);
+        if (atx) {
+            headings.push({
+                level: atx[1].length,
+                text: atx[2].trim(),
+                lineIndex: index,
+            });
+            continue;
+        }
+        const next = lines[index + 1]?.trim();
+        if (line.trim().length > 0 && next && /^(=+|-+)\s*$/.test(next)) {
+            headings.push({
+                level: next.startsWith("=") ? 1 : 2,
+                text: line.trim(),
+                lineIndex: index,
+            });
+            index++;
+        }
+    }
+    return headings;
+}
+function buildSections(headings, lines) {
+    const stack = [];
+    const sections = [];
+    const firstHeading = headings[0];
+    if (firstHeading.lineIndex > 0 &&
+        lines.slice(0, firstHeading.lineIndex).join("\n").trim().length > 0) {
+        sections.push({
+            heading: null,
+            startIndex: 0,
+            endIndex: firstHeading.lineIndex - 1,
+            breadcrumb: [],
+        });
+    }
+    for (let index = 0; index < headings.length; index++) {
+        const heading = headings[index];
+        while (stack.length > 0 && stack[stack.length - 1].level >= heading.level) {
+            stack.pop();
+        }
+        sections.push({
+            heading,
+            startIndex: heading.lineIndex,
+            endIndex: index + 1 < headings.length
+                ? headings[index + 1].lineIndex - 1
+                : lines.length - 1,
+            breadcrumb: stack.map((item) => item.text),
+        });
+        stack.push(heading);
+    }
+    return sections;
+}
+function splitMarkdownSection(input) {
+    const windows = [];
+    let startIndex = input.section.startIndex;
+    while (startIndex <= input.section.endIndex) {
+        if (input.lines[startIndex].length + 1 > input.maxChars) {
+            windows.push(...splitLongLine(input.lines[startIndex], startIndex, input.lineOffsets[startIndex], input.maxChars));
+            startIndex++;
+            continue;
+        }
+        let endIndex = startIndex;
+        let usedChars = 0;
+        while (endIndex <= input.section.endIndex) {
+            const lineLength = input.lines[endIndex].length + 1;
+            if (usedChars + lineLength > input.maxChars && endIndex > startIndex) {
+                break;
+            }
+            usedChars += lineLength;
+            endIndex++;
+        }
+        if (endIndex <= input.section.endIndex && endIndex - startIndex > 1) {
+            endIndex = chooseMarkdownBreak(input.lines, input.fenceLines, startIndex, endIndex);
+        }
+        windows.push(linesToWindow(input.lines, input.lineOffsets, startIndex, endIndex - 1));
+        if (endIndex > input.section.endIndex) {
+            break;
+        }
+        const overlapLines = computeMarkdownOverlapLines(input.lines, startIndex, endIndex, input.overlapChars);
+        const nextStart = endIndex - overlapLines;
+        startIndex = nextStart > startIndex ? nextStart : endIndex;
+    }
+    return windows.filter((window) => window.text.trim().length > 0);
+}
+function chooseMarkdownBreak(lines, fenceLines, startIndex, endIndex) {
+    const minBreak = startIndex + Math.max(1, Math.floor((endIndex - startIndex) * 0.7));
+    let bestBreak = endIndex;
+    let bestScore = markdownBreakScore(lines, fenceLines, endIndex);
+    for (let index = minBreak; index <= endIndex; index++) {
+        const score = markdownBreakScore(lines, fenceLines, index);
+        if (score > bestScore) {
+            bestBreak = index;
+            bestScore = score;
+        }
+    }
+    return bestBreak;
+}
+function markdownBreakScore(lines, fenceLines, breakIndex) {
+    if (breakIndex <= 0 || breakIndex >= lines.length || fenceLines[breakIndex]) {
+        return 0;
+    }
+    const current = lines[breakIndex].trim();
+    const previous = lines[breakIndex - 1].trim();
+    if (/^#{1,6}\s+/.test(current)) {
+        return 100;
+    }
+    if (previous === "" && current === "") {
+        return 70;
+    }
+    if (previous === "") {
+        return 60;
+    }
+    if (/^([-*+]|\d+\.)\s+/.test(current)) {
+        return 35;
+    }
+    if (/^>\s+/.test(current)) {
+        return 25;
+    }
+    return 10;
+}
+function splitLongLine(line, lineIndex, lineOffset, maxChars) {
+    const windows = [];
+    let offset = 0;
+    while (offset < line.length) {
+        const remaining = line.length - offset;
+        const sliceLength = remaining <= maxChars
+            ? remaining
+            : findLineCut(line.slice(offset), maxChars);
+        const text = line.slice(offset, offset + sliceLength);
+        windows.push({
+            text,
+            range: {
+                kind: "text",
+                startLine: lineIndex + 1,
+                endLine: lineIndex + 1,
+                startOffset: lineOffset + offset,
+                endOffset: lineOffset + offset + text.length,
+            },
+        });
+        offset += sliceLength;
+    }
+    return windows;
+}
+function linesToWindow(lines, lineOffsets, startIndex, endIndex) {
+    return {
+        text: lines.slice(startIndex, endIndex + 1).join("\n"),
+        range: {
+            kind: "text",
+            startLine: startIndex + 1,
+            endLine: endIndex + 1,
+            startOffset: lineOffsets[startIndex],
+            endOffset: lineOffsets[endIndex] + lines[endIndex].length,
+        },
+    };
+}
+function computeLineOffsets(lines) {
+    const offsets = [];
+    let offset = 0;
+    for (const line of lines) {
+        offsets.push(offset);
+        offset += line.length + 1;
+    }
+    return offsets;
+}
+function computeFenceLines(lines) {
+    const inFence = new Array(lines.length).fill(false);
+    let fence = null;
+    for (let index = 0; index < lines.length; index++) {
+        const trimmed = lines[index].trimStart();
+        if (fence) {
+            inFence[index] = true;
+            if (trimmed.startsWith(fence)) {
+                fence = null;
+            }
+            continue;
+        }
+        if (trimmed.startsWith("```")) {
+            fence = "```";
+        }
+        else if (trimmed.startsWith("~~~")) {
+            fence = "~~~";
+        }
+    }
+    return inFence;
+}
+function computeMarkdownOverlapLines(lines, startIndex, endIndex, overlapChars) {
+    if (overlapChars <= 0) {
+        return 0;
+    }
+    let chars = 0;
+    let count = 0;
+    for (let index = endIndex - 1; index > startIndex; index--) {
+        chars += lines[index].length + 1;
+        if (chars > overlapChars) {
+            break;
+        }
+        count++;
+    }
+    return Math.min(count, Math.floor((endIndex - startIndex) / 2));
+}
+function findLineCut(line, maxChars) {
+    const minPosition = Math.floor(maxChars * 0.7);
+    let bestPosition = -1;
+    let bestScore = 0;
+    for (let index = minPosition; index < maxChars; index++) {
+        const character = line[index];
+        let score = 0;
+        if (character === "." || character === "!" || character === "?") {
+            score = 4;
+        }
+        else if (character === "," || character === ";" || character === ":") {
+            score = 3;
+        }
+        else if (character === " " || character === "\t") {
+            score = 2;
+        }
+        else if (character === "-" || character === "/" || character === "\\") {
+            score = 1;
+        }
+        if (score >= bestScore && score > 0) {
+            bestScore = score;
+            bestPosition = index + 1;
+        }
+    }
+    return bestPosition > 0 ? bestPosition : maxChars;
+}
+function markdownMetadata(section) {
+    return {
+        kind: "markdown",
+        heading: section.heading?.text ?? null,
+        level: section.heading?.level ?? null,
+        scope: section.breadcrumb.length > 0 ? section.breadcrumb.join("::") : null,
+    };
+}
+//# sourceMappingURL=extractor.js.map

package/dist/engine/extraction/markdown/extractor.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"extractor.js","sourceRoot":"","sources":["../../../../src/engine/extraction/markdown/extractor.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,WAAW,EAAE,MAAM,iBAAiB,CAAC;AAM9C,OAAO,EAAE,YAAY,EAAE,MAAM,WAAW,CAAC;AACzC,OAAO,EAAE,kBAAkB,EAAgC,MAAM,cAAc,CAAC;AAChF,OAAO,EAAE,yBAAyB,EAAE,MAAM,sBAAsB,CAAC;AAEjE,OAAO,EAAE,uBAAuB,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAE/E,MAAM,4BAA4B,GAAG,IAAI,CAAC;AAC1C,MAAM,oCAAoC,GAAG,GAAG,CAAC;AAoBjD,MAAM,OAAO,iBAAiB;IAC5B,KAAK,CAAC,OAAO,CACX,MAAc,EACd,UAAwB,EAAE;QAE1B,IAAI,MAAM,CAAC,IAAI,KAAK,MAAM,IAAI,MAAM,CAAC,IAAI,CAAC,MAAM,KAAK,UAAU,EAAE,CAAC;YAChE,OAAO,EAAE,CAAC;QACZ,CAAC;QAED,kBAAkB,CAAC,MAAM,CAAC,CAAC;QAC3B,MAAM,YAAY,GAAG,2BAA2B,CAAC,OAAO,CAAC,CAAC;QAE1D,MAAM,KAAK,GAAG,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QACtC,MAAM,QAAQ,GAAG,YAAY,CAAC,KAAK,CAAC,CAAC;QACrC,IAAI,QAAQ,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YAC1B,OAAO,IAAI,CAAC,QAAQ,CAAC,MAAM,EAAE,YAAY,CAAC,CAAC;QAC7C,CAAC;QAED,MAAM,WAAW,GAAG,kBAAkB,CAAC,KAAK,CAAC,CAAC;QAC9C,MAAM,UAAU,GAAG,iBAAiB,CAAC,KAAK,CAAC,CAAC;QAC5C,MAAM,QAAQ,GAAG,aAAa,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC;QAChD,MAAM,SAAS,GAAqB,EAAE,CAAC;QAEvC,KAAK,MAAM,OAAO,IAAI,QAAQ,EAAE,CAAC;YAC/B,MAAM,QAAQ,GAAG,gBAAgB,CAAC,OAAO,CAAC,CAAC;YAC3C,MAAM,mBAAmB,GAAG,uBAAuB,CACjD,YAAY,EACZ,QAAQ,CACT,CAAC;YACF,MAAM,OAAO,GAAG,oBAAoB,CAAC;gBACnC,KAAK;gBACL,WAAW;gBACX,UAAU;gBACV,OAAO;gBACP,QAAQ,EAAE,mBAAmB,CAAC,aAAa;gBAC3C,YAAY,EAAE,mBAAmB,CAAC,iBAAiB;aACpD,CAAC,CAAC;YAEH,IAAI,OAAO,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBACvB,MAAM,EAAE,GAAG,YAAY,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE,EAAE,SAAS,CAAC,MAAM,CAAC,CAAC;gBAC1D,SAAS,CAAC,IAAI,CAAC;oBACb,EAAE;oBACF,KAAK,EAAE,EAAE;oBACT,MAAM,EAAE,MAAM,CAAC,IAAI,CAAC,EAAE;oBACtB,KAAK,EAAE,aAAa,CAClB,KAAK,EACL,WAAW,EACX,OAAO,CAAC,UAAU,EAClB,OAAO,CAAC,QAAQ,CACjB,CAAC,KAAK;oBACP,OAAO,EAAE;wBACP,IAAI,EAAE,MAAM;wBACZ,IAAI,EAAE,cAAc,CAClB,eAAe,CAAC,QAAQ,CAAC,EACzB,mBAAmB,CAAC,aAAa,CAClC;qBACF;oBACD,QAAQ;iBACT,CAAC,CAAC;gBAEH,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE,CAAC;oBAC7B,SAAS,CAAC,IAAI,CACZ,wBAAwB,CACtB,MAAM,EACN,QAAQ,EACR,MAAM,EACN,SAAS,CAAC,MAAM,EAChB,EAAE,CACH,CACF,CAAC;gBACJ,CAAC;gBAED,SAAS;YACX,CAAC;YAED,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE,CAAC;gBAC7B,SAAS,CAAC,IAAI,CAAC;oBACb,EAAE,EAAE,YAAY,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE,EAAE,SAAS,CAAC,MAAM,CAAC;oBAClD,MAAM,EAAE,MAAM,CAAC,IAAI,CAAC,EAAE;oBACtB,KAAK,EAAE,MAAM,CAAC,KAAK;oBACnB,OAAO,EAAE;wBACP,IAAI,EAAE,MAAM;wBACZ,IAAI,EAAE,MAAM,CAAC,IAAI;qBAClB;oBACD,QAAQ;iBACT,CAAC,CAAC;YACL,CAAC;QACH,CAAC;QAED,OAAO,SAAS,CAAC,MAAM,GAAG,CAAC;YACzB,CAAC,CAAC,SAAS;YACX,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,MAAM,EAAE,YAAY,CAAC,CAAC;IAC1C,CAAC;IAEO,QAAQ,CACd,MAAkB,EAClB,OAA+B;QAE/B,OAAO,yBAAyB,CAC9B,MAAM,EACN,OAAO,CAAC,aAAa,EACrB,OAAO,CAAC,iBAAiB,CAC1B,CAAC;IACJ,CAAC;CACF;AAED,SAAS,2BAA2B,CAClC,OAAqB;IAErB,MAAM,aAAa,GAAG,OAAO,CAAC,aAAa,IAAI,4BAA4B,CAAC;IAC5E,MAAM,iBAAiB,GACrB,OAAO,CAAC,iBAAiB,IAAI,oCAAoC,CAAC;IAEpE,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,aAAa,CAAC,IAAI,aAAa,IAAI,CAAC,EAAE,CAAC;QAC3D,MAAM,IAAI,WAAW,CACnB,2DAA2D,EAC3D;YACE,IAAI,EAAE,yDAAyD;YAC/D,OAAO,EAAE,iBAAiB,aAAa,EAAE;SAC1C,CACF,CAAC;IACJ,CAAC;IAED,IACE,CAAC,MAAM,CAAC,SAAS,CAAC,iBAAiB,CAAC;QACpC,iBAAiB,GAAG,CAAC;QACrB,iBAAiB,IAAI,aAAa,EAClC,CAAC;QACD,MAAM,IAAI,WAAW,CACnB,mEAAmE,EACnE;YACE,IAAI,EAAE,4DAA4D;YAClE,OAAO,EAAE,iBAAiB,aAAa,sBAAsB,iBAAiB,EAAE;SACjF,CACF,CAAC;IACJ,CAAC;IAED,OAAO,EAAE,aAAa,EAAE,iBAAiB,EAAE,CAAC;AAC9C,CAAC;AAED,SAAS,wBAAwB,CAC/B,MAAkB,EAClB,QAAgC,EAChC,MAAsB,EACtB,KAAa,EACb,KAAa;IAEb,OAAO;QACL,EAAE,EAAE,YAAY,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE,EAAE,KAAK,CAAC;QACvC,KAAK;QACL,MAAM,EAAE,MAAM,CAAC,IAAI,CAAC,EAAE;QACtB,KAAK,EAAE,MAAM,CAAC,KAAK;QACnB,OAAO,EAAE;YACP,IAAI,EAAE,MAAM;YACZ,IAAI,EAAE,MAAM,CAAC,IAAI;SAClB;QACD,QAAQ;KACT,CAAC;AACJ,CAAC;AAED,SAAS,eAAe,CAAC,QAAgC;IACvD,OAAO,QAAQ,CAAC,OAAO,IAAI,kBAAkB,CAAC;AAChD,CAAC;AAED,SAAS,YAAY,CAAC,KAAwB;IAC5C,MAAM,QAAQ,GAAc,EAAE,CAAC;IAC/B,IAAI,KAAK,GAAyB,IAAI,CAAC;IAEvC,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,GAAG,KAAK,CAAC,MAAM,EAAE,KAAK,EAAE,EAAE,CAAC;QAClD,MAAM,IAAI,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC;QAC1B,MAAM,OAAO,GAAG,IAAI,CAAC,SAAS,EAAE,CAAC;QAEjC,IAAI,KAAK,EAAE,CAAC;YACV,IAAI,OAAO,CAAC,UAAU,CAAC,KAAK,CAAC,EAAE,CAAC;gBAC9B,KAAK,GAAG,IAAI,CAAC;YACf,CAAC;YACD,SAAS;QACX,CAAC;QAED,IAAI,OAAO,CAAC,UAAU,CAAC,KAAK,CAAC,EAAE,CAAC;YAC9B,KAAK,GAAG,KAAK,CAAC;YACd,SAAS;QACX,CAAC;QAED,IAAI,OAAO,CAAC,UAAU,CAAC,KAAK,CAAC,EAAE,CAAC;YAC9B,KAAK,GAAG,KAAK,CAAC;YACd,SAAS;QACX,CAAC;QAED,MAAM,GAAG,GAAG,IAAI,CAAC,KAAK,CAAC,4BAA4B,CAAC,CAAC;QACrD,IAAI,GAAG,EAAE,CAAC;YACR,QAAQ,CAAC,IAAI,CAAC;gBACZ,KAAK,EAAE,GAAG,CAAC,CAAC,CAAC,CAAC,MAAM;gBACpB,IAAI,EAAE,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE;gBACnB,SAAS,EAAE,KAAK;aACjB,CAAC,CAAC;YACH,SAAS;QACX,CAAC;QAED,MAAM,IAAI,GAAG,KAAK,CAAC,KAAK,GAAG,CAAC,CAAC,EAAE,IAAI,EAAE,CAAC;QACtC,IAAI,IAAI,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,IAAI,IAAI,IAAI,cAAc,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YAChE,QAAQ,CAAC,IAAI,CAAC;gBACZ,KAAK,EAAE,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;gBACnC,IAAI,EAAE,IAAI,CAAC,IAAI,EAAE;gBACjB,SAAS,EAAE,KAAK;aACjB,CAAC,CAAC;YACH,KAAK,EAAE,CAAC;QACV,CAAC;IACH,CAAC;IAED,OAAO,QAAQ,CAAC;AAClB,CAAC;AAED,SAAS,aAAa,CACpB,QAA4B,EAC5B,KAAwB;IAExB,MAAM,KAAK,GAAc,EAAE,CAAC;IAC5B,MAAM,QAAQ,GAAc,EAAE,CAAC;IAC/B,MAAM,YAAY,GAAG,QAAQ,CAAC,CAAC,CAAC,CAAC;IAEjC,IACE,YAAY,CAAC,SAAS,GAAG,CAAC;QAC1B,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,YAAY,CAAC,SAAS,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,EACnE,CAAC;QACD,QAAQ,CAAC,IAAI,CAAC;YACZ,OAAO,EAAE,IAAI;YACb,UAAU,EAAE,CAAC;YACb,QAAQ,EAAE,YAAY,CAAC,SAAS,GAAG,CAAC;YACpC,UAAU,EAAE,EAAE;SACf,CAAC,CAAC;IACL,CAAC;IAED,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,GAAG,QAAQ,CAAC,MAAM,EAAE,KAAK,EAAE,EAAE,CAAC;QACrD,MAAM,OAAO,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAC;QAEhC,OAAO,KAAK,CAAC,MAAM,GAAG,CAAC,IAAI,KAAK,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,KAAK,IAAI,OAAO,CAAC,KAAK,EAAE,CAAC;YAC1E,KAAK,CAAC,GAAG,EAAE,CAAC;QACd,CAAC;QAED,QAAQ,CAAC,IAAI,CAAC;YACZ,OAAO;YACP,UAAU,EAAE,OAAO,CAAC,SAAS;YAC7B,QAAQ,EACN,KAAK,GAAG,CAAC,GAAG,QAAQ,CAAC,MAAM;gBACzB,CAAC,CAAC,QAAQ,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC,SAAS,GAAG,CAAC;gBACnC,CAAC,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC;YACtB,UAAU,EAAE,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC;SAC3C,CAAC,CAAC;QACH,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IACtB,CAAC;IAED,OAAO,QAAQ,CAAC;AAClB,CAAC;AAED,SAAS,oBAAoB,CAAC,KAO7B;IACC,MAAM,OAAO,GAAqB,EAAE,CAAC;IACrC,IAAI,UAAU,GAAG,KAAK,CAAC,OAAO,CAAC,UAAU,CAAC;IAE1C,OAAO,UAAU,IAAI,KAAK,CAAC,OAAO,CAAC,QAAQ,EAAE,CAAC;QAC5C,IAAI,KAAK,CAAC,KAAK,CAAC,UAAU,CAAC,CAAC,MAAM,GAAG,CAAC,GAAG,KAAK,CAAC,QAAQ,EAAE,CAAC;YACxD,OAAO,CAAC,IAAI,CACV,GAAG,aAAa,CACd,KAAK,CAAC,KAAK,CAAC,UAAU,CAAC,EACvB,UAAU,EACV,KAAK,CAAC,WAAW,CAAC,UAAU,CAAC,EAC7B,KAAK,CAAC,QAAQ,CACf,CACF,CAAC;YACF,UAAU,EAAE,CAAC;YACb,SAAS;QACX,CAAC;QAED,IAAI,QAAQ,GAAG,UAAU,CAAC;QAC1B,IAAI,SAAS,GAAG,CAAC,CAAC;QAElB,OAAO,QAAQ,IAAI,KAAK,CAAC,OAAO,CAAC,QAAQ,EAAE,CAAC;YAC1C,MAAM,UAAU,GAAG,KAAK,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC;YACpD,IAAI,SAAS,GAAG,UAAU,GAAG,KAAK,CAAC,QAAQ,IAAI,QAAQ,GAAG,UAAU,EAAE,CAAC;gBACrE,MAAM;YACR,CAAC;YACD,SAAS,IAAI,UAAU,CAAC;YACxB,QAAQ,EAAE,CAAC;QACb,CAAC;QAED,IAAI,QAAQ,IAAI,KAAK,CAAC,OAAO,CAAC,QAAQ,IAAI,QAAQ,GAAG,UAAU,GAAG,CAAC,EAAE,CAAC;YACpE,QAAQ,GAAG,mBAAmB,CAC5B,KAAK,CAAC,KAAK,EACX,KAAK,CAAC,UAAU,EAChB,UAAU,EACV,QAAQ,CACT,CAAC;QACJ,CAAC;QAED,OAAO,CAAC,IAAI,CACV,aAAa,CAAC,KAAK,CAAC,KAAK,EAAE,KAAK,CAAC,WAAW,EAAE,UAAU,EAAE,QAAQ,GAAG,CAAC,CAAC,CACxE,CAAC;QAEF,IAAI,QAAQ,GAAG,KAAK,CAAC,OAAO,CAAC,QAAQ,EAAE,CAAC;YACtC,MAAM;QACR,CAAC;QAED,MAAM,YAAY,GAAG,2BAA2B,CAC9C,KAAK,CAAC,KAAK,EACX,UAAU,EACV,QAAQ,EACR,KAAK,CAAC,YAAY,CACnB,CAAC;QACF,MAAM,SAAS,GAAG,QAAQ,GAAG,YAAY,CAAC;QAC1C,UAAU,GAAG,SAAS,GAAG,UAAU,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,QAAQ,CAAC;IAC7D,CAAC;IAED,OAAO,OAAO,CAAC,MAAM,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;AACnE,CAAC;AAED,SAAS,mBAAmB,CAC1B,KAAwB,EACxB,UAA8B,EAC9B,UAAkB,EAClB,QAAgB;IAEhB,MAAM,QAAQ,GACZ,UAAU,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,CAAC,QAAQ,GAAG,UAAU,CAAC,GAAG,GAAG,CAAC,CAAC,CAAC;IACtE,IAAI,SAAS,GAAG,QAAQ,CAAC;IACzB,IAAI,SAAS,GAAG,kBAAkB,CAAC,KAAK,EAAE,UAAU,EAAE,QAAQ,CAAC,CAAC;IAEhE,KAAK,IAAI,KAAK,GAAG,QAAQ,EAAE,KAAK,IAAI,QAAQ,EAAE,KAAK,EAAE,EAAE,CAAC;QACtD,MAAM,KAAK,GAAG,kBAAkB,CAAC,KAAK,EAAE,UAAU,EAAE,KAAK,CAAC,CAAC;QAC3D,IAAI,KAAK,GAAG,SAAS,EAAE,CAAC;YACtB,SAAS,GAAG,KAAK,CAAC;YAClB,SAAS,GAAG,KAAK,CAAC;QACpB,CAAC;IACH,CAAC;IAED,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,SAAS,kBAAkB,CACzB,KAAwB,EACxB,UAA8B,EAC9B,UAAkB;IAElB,IAAI,UAAU,IAAI,CAAC,IAAI,UAAU,IAAI,KAAK,CAAC,MAAM,IAAI,UAAU,CAAC,UAAU,CAAC,EAAE,CAAC;QAC5E,OAAO,CAAC,CAAC;IACX,CAAC;IAED,MAAM,OAAO,GAAG,KAAK,CAAC,UAAU,CAAC,CAAC,IAAI,EAAE,CAAC;IACzC,MAAM,QAAQ,GAAG,KAAK,CAAC,UAAU,GAAG,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;IAE9C,IAAI,YAAY,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,CAAC;QAC/B,OAAO,GAAG,CAAC;IACb,CAAC;IAED,IAAI,QAAQ,KAAK,EAAE,IAAI,OAAO,KAAK,EAAE,EAAE,CAAC;QACtC,OAAO,EAAE,CAAC;IACZ,CAAC;IAED,IAAI,QAAQ,KAAK,EAAE,EAAE,CAAC;QACpB,OAAO,EAAE,CAAC;IACZ,CAAC;IAED,IAAI,mBAAmB,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,CAAC;QACtC,OAAO,EAAE,CAAC;IACZ,CAAC;IAED,IAAI,OAAO,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,CAAC;QAC1B,OAAO,EAAE,CAAC;IACZ,CAAC;IAED,OAAO,EAAE,CAAC;AACZ,CAAC;AAED,SAAS,aAAa,CACpB,IAAY,EACZ,SAAiB,EACjB,UAAkB,EAClB,QAAgB;IAEhB,MAAM,OAAO,GAAqB,EAAE,CAAC;IACrC,IAAI,MAAM,GAAG,CAAC,CAAC;IAEf,OAAO,MAAM,GAAG,IAAI,CAAC,MAAM,EAAE,CAAC;QAC5B,MAAM,SAAS,GAAG,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC;QACvC,MAAM,WAAW,GACf,SAAS,IAAI,QAAQ;YACnB,CAAC,CAAC,SAAS;YACX,CAAC,CAAC,WAAW,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,EAAE,QAAQ,CAAC,CAAC;QAChD,MAAM,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,MAAM,GAAG,WAAW,CAAC,CAAC;QAEtD,OAAO,CAAC,IAAI,CAAC;YACX,IAAI;YACJ,KAAK,EAAE;gBACL,IAAI,EAAE,MAAM;gBACZ,SAAS,EAAE,SAAS,GAAG,CAAC;gBACxB,OAAO,EAAE,SAAS,GAAG,CAAC;gBACtB,WAAW,EAAE,UAAU,GAAG,MAAM;gBAChC,SAAS,EAAE,UAAU,GAAG,MAAM,GAAG,IAAI,CAAC,MAAM;aAC7C;SACF,CAAC,CAAC;QACH,MAAM,IAAI,WAAW,CAAC;IACxB,CAAC;IAED,OAAO,OAAO,CAAC;AACjB,CAAC;AAED,SAAS,aAAa,CACpB,KAAwB,EACxB,WAA8B,EAC9B,UAAkB,EAClB,QAAgB;IAEhB,OAAO;QACL,IAAI,EAAE,KAAK,CAAC,KAAK,CAAC,UAAU,EAAE,QAAQ,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC;QACtD,KAAK,EAAE;YACL,IAAI,EAAE,MAAM;YACZ,SAAS,EAAE,UAAU,GAAG,CAAC;YACzB,OAAO,EAAE,QAAQ,GAAG,CAAC;YACrB,WAAW,EAAE,WAAW,CAAC,UAAU,CAAC;YACpC,SAAS,EAAE,WAAW,CAAC,QAAQ,CAAC,GAAG,KAAK,CAAC,QAAQ,CAAC,CAAC,MAAM;SAC1D;KACF,CAAC;AACJ,CAAC;AAED,SAAS,kBAAkB,CAAC,KAAwB;IAClD,MAAM,OAAO,GAAa,EAAE,CAAC;IAC7B,IAAI,MAAM,GAAG,CAAC,CAAC;IAEf,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QACzB,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QACrB,MAAM,IAAI,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC;IAC5B,CAAC;IAED,OAAO,OAAO,CAAC;AACjB,CAAC;AAED,SAAS,iBAAiB,CAAC,KAAwB;IACjD,MAAM,OAAO,GAAG,IAAI,KAAK,CAAU,KAAK,CAAC,MAAM,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;IAC7D,IAAI,KAAK,GAAyB,IAAI,CAAC;IAEvC,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,GAAG,KAAK,CAAC,MAAM,EAAE,KAAK,EAAE,EAAE,CAAC;QAClD,MAAM,OAAO,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC,SAAS,EAAE,CAAC;QAEzC,IAAI,KAAK,EAAE,CAAC;YACV,OAAO,CAAC,KAAK,CAAC,GAAG,IAAI,CAAC;YACtB,IAAI,OAAO,CAAC,UAAU,CAAC,KAAK,CAAC,EAAE,CAAC;gBAC9B,KAAK,GAAG,IAAI,CAAC;YACf,CAAC;YACD,SAAS;QACX,CAAC;QAED,IAAI,OAAO,CAAC,UAAU,CAAC,KAAK,CAAC,EAAE,CAAC;YAC9B,KAAK,GAAG,KAAK,CAAC;QAChB,CAAC;aAAM,IAAI,OAAO,CAAC,UAAU,CAAC,KAAK,CAAC,EAAE,CAAC;YACrC,KAAK,GAAG,KAAK,CAAC;QAChB,CAAC;IACH,CAAC;IAED,OAAO,OAAO,CAAC;AACjB,CAAC;AAED,SAAS,2BAA2B,CAClC,KAAwB,EACxB,UAAkB,EAClB,QAAgB,EAChB,YAAoB;IAEpB,IAAI,YAAY,IAAI,CAAC,EAAE,CAAC;QACtB,OAAO,CAAC,CAAC;IACX,CAAC;IAED,IAAI,KAAK,GAAG,CAAC,CAAC;IACd,IAAI,KAAK,GAAG,CAAC,CAAC;IAEd,KAAK,IAAI,KAAK,GAAG,QAAQ,GAAG,CAAC,EAAE,KAAK,GAAG,UAAU,EAAE,KAAK,EAAE,EAAE,CAAC;QAC3D,KAAK,IAAI,KAAK,CAAC,KAAK,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC;QACjC,IAAI,KAAK,GAAG,YAAY,EAAE,CAAC;YACzB,MAAM;QACR,CAAC;QACD,KAAK,EAAE,CAAC;IACV,CAAC;IAED,OAAO,IAAI,CAAC,GAAG,CAAC,KAAK,EAAE,IAAI,CAAC,KAAK,CAAC,CAAC,QAAQ,GAAG,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;AAClE,CAAC;AAED,SAAS,WAAW,CAAC,IAAY,EAAE,QAAgB;IACjD,MAAM,WAAW,GAAG,IAAI,CAAC,KAAK,CAAC,QAAQ,GAAG,GAAG,CAAC,CAAC;IAC/C,IAAI,YAAY,GAAG,CAAC,CAAC,CAAC;IACtB,IAAI,SAAS,GAAG,CAAC,CAAC;IAElB,KAAK,IAAI,KAAK,GAAG,WAAW,EAAE,KAAK,GAAG,QAAQ,EAAE,KAAK,EAAE,EAAE,CAAC;QACxD,MAAM,SAAS,GAAG,IAAI,CAAC,KAAK,CAAC,CAAC;QAC9B,IAAI,KAAK,GAAG,CAAC,CAAC;QAEd,IAAI,SAAS,KAAK,GAAG,IAAI,SAAS,KAAK,GAAG,IAAI,SAAS,KAAK,GAAG,EAAE,CAAC;YAChE,KAAK,GAAG,CAAC,CAAC;QACZ,CAAC;aAAM,IAAI,SAAS,KAAK,GAAG,IAAI,SAAS,KAAK,GAAG,IAAI,SAAS,KAAK,GAAG,EAAE,CAAC;YACvE,KAAK,GAAG,CAAC,CAAC;QACZ,CAAC;aAAM,IAAI,SAAS,KAAK,GAAG,IAAI,SAAS,KAAK,IAAI,EAAE,CAAC;YACnD,KAAK,GAAG,CAAC,CAAC;QACZ,CAAC;aAAM,IAAI,SAAS,KAAK,GAAG,IAAI,SAAS,KAAK,GAAG,IAAI,SAAS,KAAK,IAAI,EAAE,CAAC;YACxE,KAAK,GAAG,CAAC,CAAC;QACZ,CAAC;QAED,IAAI,KAAK,IAAI,SAAS,IAAI,KAAK,GAAG,CAAC,EAAE,CAAC;YACpC,SAAS,GAAG,KAAK,CAAC;YAClB,YAAY,GAAG,KAAK,GAAG,CAAC,CAAC;QAC3B,CAAC;IACH,CAAC;IAED,OAAO,YAAY,GAAG,CAAC,CAAC,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,QAAQ,CAAC;AACpD,CAAC;AAED,SAAS,gBAAgB,CAAC,OAAgB;IACxC,OAAO;QACL,IAAI,EAAE,UAAU;QAChB,OAAO,EAAE,OAAO,CAAC,OAAO,EAAE,IAAI,IAAI,IAAI;QACtC,KAAK,EAAE,OAAO,CAAC,OAAO,EAAE,KAAK,IAAI,IAAI;QACrC,KAAK,EAAE,OAAO,CAAC,UAAU,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI;KAC5E,CAAC;AACJ,CAAC"}

package/dist/engine/extraction/runtime.d.ts

@@ -0,0 +1,9 @@
+import type { Content, EntityFragment } from "../types.js";
+import type { Source } from "./source.js";
+import type { ChunkOptions } from "./types.js";
+export declare function extract(source: Source, options?: ChunkOptions): Promise<EntityFragment[]>;
+export type IndexingExtractionFragment = {
+    fragment: EntityFragment;
+    embeddingSource?: Content;
+};
+export declare function extractForIndexing(source: Source, options?: ChunkOptions): Promise<IndexingExtractionFragment[]>;

package/dist/engine/extraction/runtime.js

@@ -0,0 +1,37 @@
+import { CodeExtractor } from "./code/extractor.js";
+import { ImageExtractor } from "./image/extractor.js";
+import { MarkdownExtractor } from "./markdown/extractor.js";
+import { TextExtractor } from "./text/extractor.js";
+const extractors = {
+    code: new CodeExtractor(),
+    image: new ImageExtractor(),
+    markdown: new MarkdownExtractor(),
+    text: new TextExtractor(),
+};
+export function extract(source, options = {}) {
+    return extractors[routeSource(source)].extract(source, options);
+}
+export async function extractForIndexing(source, options = {}) {
+    if (routeSource(source) === "code") {
+        return (await extractors.code.extractForIndexing(source, options)).map(({ fragment, embeddingText }) => ({
+            fragment,
+            ...(embeddingText === undefined
+                ? {}
+                : { embeddingSource: { kind: "text", text: embeddingText } }),
+        }));
+    }
+    return (await extractors[routeSource(source)].extract(source, options)).map((fragment) => ({ fragment }));
+}
+function routeSource(source) {
+    if (source.kind === "image") {
+        return "image";
+    }
+    if (source.file.kind === "code") {
+        return "code";
+    }
+    if (source.file.format === "markdown") {
+        return "markdown";
+    }
+    return "text";
+}
+//# sourceMappingURL=runtime.js.map

package/dist/engine/extraction/runtime.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"runtime.js","sourceRoot":"","sources":["../../../src/engine/extraction/runtime.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,aAAa,EAAE,MAAM,qBAAqB,CAAC;AACpD,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AACtD,OAAO,EAAE,iBAAiB,EAAE,MAAM,yBAAyB,CAAC;AAE5D,OAAO,EAAE,aAAa,EAAE,MAAM,qBAAqB,CAAC;AASpD,MAAM,UAAU,GAAG;IACjB,IAAI,EAAE,IAAI,aAAa,EAAE;IACzB,KAAK,EAAE,IAAI,cAAc,EAAE;IAC3B,QAAQ,EAAE,IAAI,iBAAiB,EAAE;IACjC,IAAI,EAAE,IAAI,aAAa,EAAE;CACwB,CAAC;AAEpD,MAAM,UAAU,OAAO,CACrB,MAAc,EACd,UAAwB,EAAE;IAE1B,OAAO,UAAU,CAAC,WAAW,CAAC,MAAM,CAAC,CAAC,CAAC,OAAO,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;AAClE,CAAC;AAOD,MAAM,CAAC,KAAK,UAAU,kBAAkB,CACtC,MAAc,EACd,UAAwB,EAAE;IAE1B,IAAI,WAAW,CAAC,MAAM,CAAC,KAAK,MAAM,EAAE,CAAC;QACnC,OAAO,CAAC,MAAM,UAAU,CAAC,IAAI,CAAC,kBAAkB,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC,CAAC,GAAG,CACpE,CAAC,EAAE,QAAQ,EAAE,aAAa,EAAE,EAAE,EAAE,CAAC,CAAC;YAChC,QAAQ;YACR,GAAG,CAAC,aAAa,KAAK,SAAS;gBAC7B,CAAC,CAAC,EAAE;gBACJ,CAAC,CAAC,EAAE,eAAe,EAAE,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,aAAa,EAAE,EAAE,CAAC;SAChE,CAAC,CACH,CAAC;IACJ,CAAC;IAED,OAAO,CAAC,MAAM,UAAU,CAAC,WAAW,CAAC,MAAM,CAAC,CAAC,CAAC,OAAO,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC,CAAC,GAAG,CACzE,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC,EAAE,QAAQ,EAAE,CAAC,CAC7B,CAAC;AACJ,CAAC;AAED,SAAS,WAAW,CAAC,MAAc;IACjC,IAAI,MAAM,CAAC,IAAI,KAAK,OAAO,EAAE,CAAC;QAC5B,OAAO,OAAO,CAAC;IACjB,CAAC;IAED,IAAI,MAAM,CAAC,IAAI,CAAC,IAAI,KAAK,MAAM,EAAE,CAAC;QAChC,OAAO,MAAM,CAAC;IAChB,CAAC;IAED,IAAI,MAAM,CAAC,IAAI,CAAC,MAAM,KAAK,UAAU,EAAE,CAAC;QACtC,OAAO,UAAU,CAAC;IACpB,CAAC;IAED,OAAO,MAAM,CAAC;AAChB,CAAC"}

package/dist/engine/extraction/source.d.ts

@@ -0,0 +1,18 @@
+import type { FileInfo, ImageFormat } from "../types.js";
+export type SourceKind = "text" | "image";
+type SourceBase = {
+    kind: SourceKind;
+    file: FileInfo;
+};
+export type TextSource = SourceBase & {
+    kind: "text";
+    text: string;
+};
+export type ImageSource = SourceBase & {
+    kind: "image";
+    data: Uint8Array;
+    format: ImageFormat;
+};
+export type Source = TextSource | ImageSource;
+export declare function validateSourceFile(source: Source): void;
+export {};

package/dist/engine/extraction/source.js

@@ -0,0 +1,22 @@
+import { EngineError } from "../errors.js";
+export function validateSourceFile(source) {
+    if (source.file.id.trim().length === 0) {
+        throw new EngineError("Extractor source requires a non-empty file id", {
+            code: "ZVEC_GREP.ENGINE.EXTRACTORS.EMPTY_FILE_ID",
+            context: `sourceKind=${source.kind}`,
+        });
+    }
+    if (source.file.absolutePath.trim().length === 0) {
+        throw new EngineError("Extractor source requires a non-empty absolute file path", {
+            code: "ZVEC_GREP.ENGINE.EXTRACTORS.EMPTY_ABSOLUTE_FILE_PATH",
+            context: `fileId=${source.file.id} sourceKind=${source.kind}`,
+        });
+    }
+    if (source.file.relativePath.trim().length === 0) {
+        throw new EngineError("Extractor source requires a non-empty relative file path", {
+            code: "ZVEC_GREP.ENGINE.EXTRACTORS.EMPTY_RELATIVE_FILE_PATH",
+            context: `fileId=${source.file.id} sourceKind=${source.kind}`,
+        });
+    }
+}
+//# sourceMappingURL=source.js.map

package/dist/engine/extraction/source.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"source.js","sourceRoot":"","sources":["../../../src/engine/extraction/source.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,WAAW,EAAE,MAAM,cAAc,CAAC;AAuB3C,MAAM,UAAU,kBAAkB,CAAC,MAAc;IAC/C,IAAI,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QACvC,MAAM,IAAI,WAAW,CAAC,+CAA+C,EAAE;YACrE,IAAI,EAAE,2CAA2C;YACjD,OAAO,EAAE,cAAc,MAAM,CAAC,IAAI,EAAE;SACrC,CAAC,CAAC;IACL,CAAC;IAED,IAAI,MAAM,CAAC,IAAI,CAAC,YAAY,CAAC,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QACjD,MAAM,IAAI,WAAW,CACnB,0DAA0D,EAC1D;YACE,IAAI,EAAE,sDAAsD;YAC5D,OAAO,EAAE,UAAU,MAAM,CAAC,IAAI,CAAC,EAAE,eAAe,MAAM,CAAC,IAAI,EAAE;SAC9D,CACF,CAAC;IACJ,CAAC;IAED,IAAI,MAAM,CAAC,IAAI,CAAC,YAAY,CAAC,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QACjD,MAAM,IAAI,WAAW,CACnB,0DAA0D,EAC1D;YACE,IAAI,EAAE,sDAAsD;YAC5D,OAAO,EAAE,UAAU,MAAM,CAAC,IAAI,CAAC,EAAE,eAAe,MAAM,CAAC,IAAI,EAAE;SAC9D,CACF,CAAC;IACJ,CAAC;AACH,CAAC"}

package/dist/engine/extraction/text/extractor.d.ts

@@ -0,0 +1,7 @@
+import type { EntityFragment } from "../../types.js";
+import { type Source, type TextSource } from "../source.js";
+import type { ChunkOptions } from "../types.js";
+export declare class TextExtractor {
+    extract(source: Source, options?: ChunkOptions): Promise<EntityFragment[]>;
+}
+export declare function extractPlainTextFragments(source: TextSource, maxChunkChars: number, chunkOverlapChars: number): EntityFragment[];

package/dist/engine/extraction/text/extractor.js

@@ -0,0 +1,168 @@
+import { EngineError } from "../../errors.js";
+import { makeEntityId } from "../ids.js";
+import { validateSourceFile } from "../source.js";
+const DEFAULT_TEXT_CHUNK_CHARS = 3600;
+const DEFAULT_TEXT_CHUNK_OVERLAP_CHARS = 540;
+export class TextExtractor {
+    async extract(source, options = {}) {
+        if (source.kind !== "text") {
+            throw new EngineError("Text extractor received a non-text source", {
+                code: "ZVEC_GREP.ENGINE.EXTRACTORS.TEXT_UNSUPPORTED_SOURCE",
+                context: `fileId=${source.file.id} sourceKind=${source.kind}`,
+            });
+        }
+        validateSourceFile(source);
+        const { maxChunkChars, chunkOverlapChars } = resolveTextChunkOptions(options);
+        return extractPlainTextFragments(source, maxChunkChars, chunkOverlapChars);
+    }
+}
+function resolveTextChunkOptions(options) {
+    const maxChunkChars = options.maxChunkChars ?? DEFAULT_TEXT_CHUNK_CHARS;
+    const chunkOverlapChars = options.chunkOverlapChars ?? DEFAULT_TEXT_CHUNK_OVERLAP_CHARS;
+    if (!Number.isInteger(maxChunkChars) || maxChunkChars <= 0) {
+        throw new EngineError("Text extractor requires a positive integer chunk size", {
+            code: "ZVEC_GREP.ENGINE.EXTRACTORS.TEXT_INVALID_CHUNK_SIZE",
+            context: `maxChunkChars=${maxChunkChars}`,
+        });
+    }
+    if (!Number.isInteger(chunkOverlapChars) ||
+        chunkOverlapChars < 0 ||
+        chunkOverlapChars >= maxChunkChars) {
+        throw new EngineError("Text extractor requires overlap to be smaller than chunk size", {
+            code: "ZVEC_GREP.ENGINE.EXTRACTORS.TEXT_INVALID_CHUNK_OVERLAP",
+            context: `maxChunkChars=${maxChunkChars} chunkOverlapChars=${chunkOverlapChars}`,
+        });
+    }
+    return { maxChunkChars, chunkOverlapChars };
+}
+export function extractPlainTextFragments(source, maxChunkChars, chunkOverlapChars) {
+    return chunkText(source.text, maxChunkChars, chunkOverlapChars).map((chunk, index) => ({
+        id: makeEntityId(source.file.id, index),
+        fileId: source.file.id,
+        range: chunk.range,
+        content: {
+            kind: "text",
+            text: chunk.text,
+        },
+    }));
+}
+function chunkText(text, maxChars, overlapChars) {
+    if (text.trim().length === 0) {
+        return [];
+    }
+    const lines = text.split("\n");
+    const lineOffsets = computeLineOffsets(lines);
+    const chunks = [];
+    let startIndex = 0;
+    while (startIndex < lines.length) {
+        if (lines[startIndex].length + 1 > maxChars) {
+            const line = lines[startIndex];
+            const lineNumber = startIndex + 1;
+            let offset = 0;
+            while (offset < line.length) {
+                const remaining = line.length - offset;
+                const sliceLength = remaining <= maxChars
+                    ? remaining
+                    : findLineCut(line.slice(offset), maxChars);
+                const slice = line.slice(offset, offset + sliceLength);
+                if (slice.trim().length > 0) {
+                    chunks.push({
+                        text: slice,
+                        range: {
+                            kind: "text",
+                            startLine: lineNumber,
+                            endLine: lineNumber,
+                            startOffset: lineOffsets[startIndex] + offset,
+                            endOffset: lineOffsets[startIndex] + offset + slice.length,
+                        },
+                    });
+                }
+                offset += sliceLength;
+            }
+            startIndex++;
+            continue;
+        }
+        let usedChars = 0;
+        let endIndex = startIndex;
+        while (endIndex < lines.length) {
+            const lineLength = lines[endIndex].length + 1;
+            if (usedChars + lineLength > maxChars && endIndex > startIndex) {
+                break;
+            }
+            usedChars += lineLength;
+            endIndex++;
+        }
+        const chunkLines = lines.slice(startIndex, endIndex);
+        const chunk = chunkLines.join("\n");
+        if (chunk.trim().length > 0) {
+            const endLineIndex = endIndex - 1;
+            chunks.push({
+                text: chunk,
+                range: {
+                    kind: "text",
+                    startLine: startIndex + 1,
+                    endLine: endIndex,
+                    startOffset: lineOffsets[startIndex],
+                    endOffset: lineOffsets[endLineIndex] + lines[endLineIndex].length,
+                },
+            });
+        }
+        if (endIndex >= lines.length) {
+            break;
+        }
+        startIndex = computeNextStartLine(lines, startIndex, endIndex, overlapChars);
+    }
+    return chunks;
+}
+function computeLineOffsets(lines) {
+    const offsets = [];
+    let offset = 0;
+    for (const line of lines) {
+        offsets.push(offset);
+        offset += line.length + 1;
+    }
+    return offsets;
+}
+function computeNextStartLine(lines, startIndex, endIndex, overlapChars) {
+    if (overlapChars === 0) {
+        return endIndex;
+    }
+    let overlapLines = 0;
+    let overlapCount = 0;
+    for (let index = endIndex - 1; index >= startIndex && overlapCount < overlapChars; index--) {
+        overlapCount += lines[index].length + 1;
+        overlapLines++;
+    }
+    const nextStart = endIndex - overlapLines;
+    return nextStart > startIndex ? nextStart : endIndex;
+}
+function findLineCut(line, maxChars) {
+    if (line.length <= maxChars) {
+        return line.length;
+    }
+    const minPosition = Math.floor(maxChars * 0.7);
+    let bestPosition = -1;
+    let bestScore = 0;
+    for (let index = minPosition; index < maxChars; index++) {
+        const character = line[index];
+        let score = 0;
+        if (character === "." || character === "!" || character === "?") {
+            score = 4;
+        }
+        else if (character === "," || character === ";" || character === ":") {
+            score = 3;
+        }
+        else if (character === " " || character === "\t") {
+            score = 2;
+        }
+        else if (character === "-" || character === "/" || character === "\\") {
+            score = 1;
+        }
+        if (score > 0 && score >= bestScore) {
+            bestScore = score;
+            bestPosition = index + 1;
+        }
+    }
+    return bestPosition > 0 ? bestPosition : maxChars;
+}
+//# sourceMappingURL=extractor.js.map