docrev 0.7.8 → 0.7.9

package/lib/build.js

@@ -213,7 +213,7 @@ export function combineSections(directory, config, options = {}) {
   // Resolve forward references (refs that appear before their anchor definition)
   // This fixes pandoc-crossref limitation with multi-file documents
   if (hasPandocCrossref()) {
-    const registry = buildRegistry(directory);
+    const registry = buildRegistry(directory, sections);
     const { text, resolved } = resolveForwardRefs(paperContent, registry);
     if (resolved.length > 0) {
       paperContent = text;
@@ -280,7 +280,8 @@ export function prepareForFormat(paperPath, format, config, options = {}) {
   let content = fs.readFileSync(paperPath, 'utf-8');
 
   // Build crossref registry for reference conversion
-  const registry = buildRegistry(directory);
+  // Pass sections from config to ensure correct file ordering
+  const registry = buildRegistry(directory, config.sections);
 
   if (format === 'pdf' || format === 'tex') {
     // Strip all annotations for clean output

package/lib/crossref.js

@@ -11,6 +11,49 @@ import * as fs from 'fs';
 import * as path from 'path';
 import YAML from 'yaml';
 
+/**
+ * Discover section files from a directory by reading config files
+ * Only returns files explicitly defined in rev.yaml or sections.yaml
+ * Returns empty array if no config found (caller should handle this)
+ *
+ * @param {string} directory
+ * @returns {string[]} Ordered list of section filenames, or empty if no config
+ */
+function discoverSectionFiles(directory) {
+  // Try rev.yaml first
+  const revYamlPath = path.join(directory, 'rev.yaml');
+  if (fs.existsSync(revYamlPath)) {
+    try {
+      const config = YAML.parse(fs.readFileSync(revYamlPath, 'utf-8'));
+      if (config.sections && Array.isArray(config.sections) && config.sections.length > 0) {
+        return config.sections.filter(f => fs.existsSync(path.join(directory, f)));
+      }
+    } catch {
+      // Ignore yaml errors, try next option
+    }
+  }
+
+  // Try sections.yaml
+  const sectionsPath = path.join(directory, 'sections.yaml');
+  if (fs.existsSync(sectionsPath)) {
+    try {
+      const config = YAML.parse(fs.readFileSync(sectionsPath, 'utf-8'));
+      if (config.sections) {
+        const sectionOrder = Object.entries(config.sections)
+          .sort((a, b) => (a[1].order ?? 999) - (b[1].order ?? 999))
+          .map(([file]) => file);
+        return sectionOrder.filter(f => fs.existsSync(path.join(directory, f)));
+      }
+    } catch {
+      // Ignore yaml errors
+    }
+  }
+
+  // No config found - return empty array
+  // Caller must handle this (either error or use explicit sections)
+  return [];
+}
+
 /**
  * Patterns for detecting hardcoded references
  * Matches complex patterns including:
@@ -208,8 +251,14 @@ export function parseReferenceList(listStr) {
  * Build a registry of figure/table labels from .md files
  * Scans for {#fig:label} and {#tbl:label} anchors
  *
+ * IMPORTANT: This function requires either explicit sections or a rev.yaml/sections.yaml config.
+ * It will NOT guess by scanning all .md files, as this leads to incorrect numbering
+ * when temporary files (paper_clean.md, etc.) exist in the directory.
+ *
  * @param {string} directory - Directory containing .md files
- * @param {string[]} [excludeFiles] - Files to exclude
+ * @param {string[]} [sections] - Array of section filenames to scan (recommended).
+ *   If not provided, reads from rev.yaml or sections.yaml.
+ *   Returns empty registry if no sections can be determined.
  * @returns {{
  *   figures: Map<string, {label: string, num: number, isSupp: boolean, file: string}>,
  *   tables: Map<string, {label: string, num: number, isSupp: boolean, file: string}>,
@@ -217,7 +266,7 @@ export function parseReferenceList(listStr) {
  *   byNumber: {fig: Map<string, string>, tbl: Map<string, string>, eq: Map<string, string>}
  * }}
  */
-export function buildRegistry(directory, excludeFiles = ['paper.md', 'README.md', 'CLAUDE.md']) {
+export function buildRegistry(directory, sections) {
   const figures = new Map();
   const tables = new Map();
   const equations = new Map();
@@ -229,32 +278,16 @@ export function buildRegistry(directory, excludeFiles = ['paper.md', 'README.md'
   let tblSuppNum = 0;
   let eqNum = 0;
 
-  // Get all .md files
-  const files = fs.readdirSync(directory).filter((f) => {
-    if (!f.endsWith('.md')) return false;
-    if (excludeFiles.some((e) => f.toLowerCase() === e.toLowerCase())) return false;
-    return true;
-  });
+  let orderedFiles;
 
-  // Sort by likely document order (use sections.yaml if available)
-  let orderedFiles = files;
-  const sectionsPath = path.join(directory, 'sections.yaml');
-  if (fs.existsSync(sectionsPath)) {
-    try {
-      const config = YAML.parse(fs.readFileSync(sectionsPath, 'utf-8'));
-      if (config.sections) {
-        const sectionOrder = Object.entries(config.sections)
-          .sort((a, b) => (a[1].order ?? 999) - (b[1].order ?? 999))
-          .map(([file]) => file);
-        orderedFiles = sectionOrder.filter((f) => files.includes(f));
-        // Add any remaining files not in sections.yaml
-        for (const f of files) {
-          if (!orderedFiles.includes(f)) orderedFiles.push(f);
-        }
-      }
-    } catch {
-      // Ignore yaml errors, use default order
-    }
+  if (Array.isArray(sections) && sections.length > 0) {
+    // Use explicitly provided section files - most reliable
+    orderedFiles = sections.filter(f => fs.existsSync(path.join(directory, f)));
+  } else {
+    // Try to determine sections from config files (rev.yaml or sections.yaml)
+    orderedFiles = discoverSectionFiles(directory);
+    // If no config found, return empty registry rather than guessing
+    // This prevents bugs from scanning wrong files
   }
 
   // Determine if a file is supplementary

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "docrev",
-  "version": "0.7.8",
+  "version": "0.7.9",
   "description": "Academic paper revision workflow: Word ↔ Markdown round-trips, DOI validation, reviewer comments",
   "type": "module",
   "types": "types/index.d.ts",