fix(heuristics): two-column flatten keeps flat-bullet awards as achievements (#131)

src/lib/heuristics/entry-blocks.test.ts

@@ -281,6 +281,45 @@ describe("parseEntryBlocks — first_line anchor (projects / date-optional secti
     };
   }
 
+  it("splits a flat bullet list (awards) into one entry per top-level bullet (#131)", () => {
+    // An achievements/awards section where every item is itself a bullet — there
+    // is no non-bullet header line for the first_line anchor to latch onto. Each
+    // bullet must become its own entry, with a marker-less year line below it
+    // (a wrapped tail at x=54 > the bullet margin x=48) folding into that entry.
+    const blocks = parseEntryBlocks(
+      sectionX([
+        { text: "• Globex Engineering Excellence,", x: 48 },
+        { text: "2021", x: 54 },
+        { text: "• Acme Innovation Prize, 2023", x: 48 },
+      ]),
+      { anchor: "first_line", collectBody: true },
+    );
+    expect(blocks).toHaveLength(2);
+    expect(blocks[0].headerLines).toEqual(["Globex Engineering Excellence"]);
+    expect(blocks[0].dates.start_date).toBe("2021");
+    expect(blocks[1].headerLines).toEqual(["Acme Innovation Prize"]);
+    expect(blocks[1].dates.start_date).toBe("2023");
+  });
+
+  it("keeps a deeper sub-bullet as the entry body, not a new entry (#131)", () => {
+    // A top-level award bullet (x=48) with a deeper-indented detail bullet
+    // (x=72) under it: the sub-bullet is the body of that one award, not a
+    // second entry.
+    const blocks = parseEntryBlocks(
+      sectionX([
+        { text: "• Employee of the Year, 2024", x: 48 },
+        { text: "• Recognized for cross-team leadership.", x: 72 },
+        { text: "• Best Demo Award, 2023", x: 48 },
+      ]),
+      { anchor: "first_line", collectBody: true },
+    );
+    expect(blocks).toHaveLength(2);
+    expect(blocks[0].headerLines).toEqual(["Employee of the Year"]);
+    expect(blocks[0].body).toContain("cross-team leadership");
+    expect(blocks[0].bulletCount).toBe(1);
+    expect(blocks[1].headerLines).toEqual(["Best Demo Award"]);
+  });
+
   it("treats an indented wrapped-bullet line as a continuation, not a new entry", () => {
     // Headers sit at the section margin (x=50); the bullet text (and thus a
     // wrapped continuation of it) is indented to x=73. The two wrap lines must

src/lib/heuristics/entry-blocks.ts

@@ -196,12 +196,88 @@ export function parseEntryBlocks(
 
   const lines = section.lines;
   const anchors = collectAnchors(lines, cfg.anchor);
-  if (anchors.length === 0) return [];
+  if (anchors.length === 0) {
+    // A `first_line` section with no anchorable header line is a flat bullet
+    // list (an awards / achievements list where every item is itself a bullet,
+    // so there is no name-led header for `collectAnchors` to latch onto). Rather
+    // than drop the whole section, fall back to anchoring on the bullets. The
+    // other anchors have no such list shape, so they keep returning [].
+    return cfg.anchor === "first_line" ? parseBulletList(lines, cfg) : [];
+  }
 
   const lookback = cfg.headerLookback ?? 0;
   return anchors.map((_, a) => buildEntryBlock(lines, anchors, a, cfg, lookback));
 }
 
+/**
+ * Fallback parser for a `first_line` section that is a flat bullet list — every
+ * entry is itself a bullet ("• Award name, 2023"), so `collectAnchors` found no
+ * non-bullet header line and returned zero anchors. Each TOP-LEVEL bullet (one
+ * sitting at the bullet-marker margin) becomes its own entry; any marker-less
+ * lines below it (a year on its own line, a wrapped award name) fold into that
+ * entry's title, and deeper-indented sub-bullets become its body.
+ *
+ * This assumes upstream column banding (`detectColumnBoundaries` in
+ * `pdf-extract.ts`) has already separated a two-column layout into single-column
+ * sections, so the lines here are one column's list — not two bullet margins
+ * interleaved. That banding is what makes a single per-section bullet margin a
+ * valid assumption (see #131).
+ */
+function parseBulletList(lines: PdfLine[], cfg: EntryBlockConfig): EntryBlock[] {
+  const markerX = bulletMarkerX(lines);
+  if (!Number.isFinite(markerX)) return [];
+  const anchors: number[] = [];
+  for (let i = 0; i < lines.length; i++) {
+    // Top-level bullets only: a deeper-indented bullet is a sub-item of the
+    // entry above it, not a new entry.
+    if (isBulletLine(lines[i]) && lines[i].x <= markerX + 2) anchors.push(i);
+  }
+  if (anchors.length === 0) return [];
+  return anchors.map((_, a) => buildBulletEntry(lines, anchors, a, cfg));
+}
+
+/**
+ * Build the single bullet-list `EntryBlock` anchored at `anchors[a]`. The entry
+ * spans to just before the next top-level bullet: the anchor bullet's text plus
+ * any marker-less continuation lines below it form the title (date stripped off,
+ * parsed onto `dates`); deeper sub-bullets form the body. Extracted from
+ * `parseBulletList` to keep each function below the cognitive-complexity bar.
+ */
+function buildBulletEntry(
+  lines: PdfLine[],
+  anchors: number[],
+  a: number,
+  cfg: EntryBlockConfig,
+): EntryBlock {
+  const anchorIdx = anchors[a];
+  const nextIdx = a + 1 < anchors.length ? anchors[a + 1] : lines.length;
+
+  const titleParts = [stripBullet(lines[anchorIdx].text)];
+  const bodyBullets: PdfLine[] = [];
+  for (let i = anchorIdx + 1; i < nextIdx; i++) {
+    if (isBulletLine(lines[i])) bodyBullets.push(lines[i]);
+    else titleParts.push(lines[i].text.trim());
+  }
+
+  const combined = titleParts.join(" ").replace(/\s+/g, " ").trim();
+  const dates = parseDateRange(combined);
+  const title = stripDateRange(combined);
+
+  const body = cfg.collectBody
+    ? bodyBullets
+        .map((l) => stripBullet(l.text))
+        .join("\n")
+        .trim() || undefined
+    : undefined;
+
+  return {
+    headerLines: title ? [title] : [],
+    dates,
+    body,
+    bulletCount: cfg.collectBody ? bodyBullets.length : 0,
+  };
+}
+
 /**
  * Build the single `EntryBlock` anchored at `anchors[a]`. The entry spans from
  * just after the previous anchor to just before the next: header lines are the

tests/fixtures/pdfs/google-docs/google-docs-skia-proxy-two-column.expected.json

@@ -20,6 +20,7 @@
       "full_name",
       "github_url",
       "given_name",
+      "heuristic_achievements",
       "linkedin_url",
       "location",
       "phone",
@@ -31,7 +32,7 @@
     "experienceCount": 3,
     "educationCount": 1,
     "projectsCount": 1,
-    "achievementsCount": 0,
+    "achievementsCount": 2,
     "rawTextCharCount": 1720,
     "pageCount": 1,
     "linkAnnotationCount": 0,

tests/fixtures/pdfs/unknown/weasyprint-cairo-two-column.expected.json

@@ -20,6 +20,7 @@
       "full_name",
       "github_url",
       "given_name",
+      "heuristic_achievements",
       "linkedin_url",
       "location",
       "phone",
@@ -31,7 +32,7 @@
     "experienceCount": 3,
     "educationCount": 1,
     "projectsCount": 1,
-    "achievementsCount": 0,
+    "achievementsCount": 2,
     "rawTextCharCount": 1722,
     "pageCount": 1,
     "linkAnnotationCount": 0,