hippo-memory 1.63.2 → 1.64.0

package/dist/capture/extract.js

@@ -1,386 +1,175 @@
-/**
- * Capture actionable items from conversation text.
- *
- * Uses heuristic pattern matching (no LLM) to extract:
- *   - Decisions ("we decided", "let's do", "going with")
- *   - Specs / requirements (bullet lists after spec/feature/plan headings)
- *   - Rules / constraints ("never", "always", "the rule is", "must")
- *   - Errors / gotchas ("error:", "bug:", "gotcha:", "watch out")
- *   - Preferences ("prefer", "use X instead of Y", "don't use")
- */
 import { duplicateKey } from '../same-text.js';
-import { isContentWorthStoring } from '../audit.js';
-// Sentence-level patterns
-//
-// Each pattern carries TWO capture groups: group 1 is the discriminating keyword (plus its
-// trailing separator, verbatim), group 2 the content after it, so a negation like "never" survives
-// ("Never use X" must not store as "use X"). `extractFromPatterns` joins group1 + a clause-bounded
-// group2 (`boundToClause`), so group 2 may reach {1,500}. The keyword keeps its own group because
-// keywords like "error:" end in a colon that is not a prose clause boundary.
-// `PREFERENCE_PATTERNS[0]` is the exception: its two groups are the content spans either side of
-// "instead of"/"over"/"not"; left as-is.
+import { assessAutomaticMemory } from '../memory-quality.js';
 const DECISION_PATTERNS = [
-    /(?:we(?:'ve| have)?|i(?:'ve| have)?|let's)\s+(decid(?:ed|e)\s+(?:to\s+)?)(.{1,500})/i,
-    /(?:let's|we(?:'ll| will| should)?)\s+((?:go with|do|use|try|build|implement|switch to)\s+)(.{1,500})/i,
-    /((?:going|went)\s+with\s+)(.{1,500})/i,
-    /((?:the plan is|plan:)\s+)(.{1,500})/i,
-    /(decision:\s*)(.{1,500})/i,
+    /\b(?:we(?:'ve| have)?|i(?:'ve| have)?|let's)\s+decid(?:ed|e)\b/i,
+    /\b(?:let's|we(?:'ll| will| should)?)\s+(?:go with|do|use|try|build|implement|switch to)\b/i,
+    /\b(?:going|went)\s+with\b/i,
+    /\b(?:the plan is|plan:|decision:)\s/i,
 ];
 const RULE_PATTERNS = [
-    /\b((?:never|always|must(?:\s+not)?|do(?:n't| not)\s+ever)\s+)(.{1,500})/i,
-    /((?:the rule is|rule:)\s*)(.{1,500})/i,
-    /((?:important|critical|remember):\s*)(.{1,500})/i,
-    /((?:make sure|ensure)\s+(?:to\s+)?)(.{1,500})/i,
+    /\b(?:never|always|must(?:\s+not)?|do(?:n't| not)\s+ever)\s/i,
+    /\b(?:the rule is|rule:|important:|critical:|remember:)\s/i,
+    /\b(?:make sure|ensure)\s/i,
 ];
 const ERROR_PATTERNS = [
-    /((?:error|bug|gotcha|watch out|careful|warning|caveat|trap):\s*)(.{1,500})/i,
-    /((?:this broke|this breaks|this will break|broke because)\s+)(.{1,500})/i,
-    /((?:the (?:issue|problem|fix) (?:is|was))\s+)(.{1,500})/i,
-    /((?:don't forget|easy to miss):\s*)(.{1,500})/i,
+    /\b(?:error|bug|gotcha|watch out|careful|warning|caveat|trap):\s/i,
+    /\b(?:this broke|this breaks|this will break|broke because)\s/i,
+    /\bthe (?:issue|problem|fix) (?:is|was)\s/i,
+    /\b(?:don't forget|easy to miss):\s/i,
 ];
-// PREFERENCE_PATTERNS[0] keeps its match[1]-only, unbounded shape; extractFromPatterns
-// checks for this exact array element.
 const PREFERENCE_PATTERNS = [
-    /(?:prefer|use)\s+(.{5,100})\s+(?:instead of|over|not)\s+(.{3,100})/i,
-    /((?:don't use|avoid|skip)\s+)(.{1,500})/i,
-    /((?:we(?:'re| are)\s+using|the stack is|we use)\s+)(.{1,500})/i,
+    /\b(?:prefer|use)\s+.+\s+(?:instead of|over|not)\s+/i,
+    /\b(?:prefer|don't use|avoid|skip)\s/i,
+    /\b(?:we(?:'re| are)\s+using|the stack is|we use)\s/i,
 ];
-// Heading patterns that signal a following list of specs/requirements
+const CATEGORIES = [
+    ['decision', DECISION_PATTERNS], ['rule', RULE_PATTERNS],
+    ['error', ERROR_PATTERNS], ['preference', PREFERENCE_PATTERNS],
+];
+const MAX_CHARS = 500;
+const BULLET = /^(?:[-*]|\d+\.)\s+(?:\[[ xX]\]\s+)?(.+)/;
+const LABEL = /^(?:decision|rule|error|bug|gotcha|important|critical|remember|plan):/i;
+const SENTENCE_END = /[.!?]["')\]]*$/;
 const SPEC_HEADING_PATTERNS = [
     /^#+\s*(?:features?|requirements?|specs?|specifications?|plan|design|architecture|interface|api|todo|tasks?|implementation|notes?)(?:\s|:|$)/i,
     /^(?:features?|requirements?|specs?|specifications?|plan|design|tasks?|implementation)(?:\s*:|$)/i,
 ];
-// ---------------------------------------------------------------------------
-// Extraction engine
-// ---------------------------------------------------------------------------
-export function splitSentences(text) {
-    // Split on sentence boundaries, keeping reasonable chunks
-    return text
-        .split(/(?<=[.!?])\s+|\n/)
-        .map((s) => s.trim())
-        .filter((s) => s.length > 5);
+// A list item may open a fence on its own line: "- ```".
+const FENCE_OPEN = /^(?:(?:[-*]|\d+\.)\s+)?(`{3,}|~{3,})(.*)$/;
+const FENCE_CLOSE = /^(`{3,}|~{3,})\s*$/;
+// Inline code and abbreviation dots never end a sentence; "etc." does when a capital follows.
+const NO_BREAK = /(`+)[^`\n]+\1|\b(?:[Ee]\.g|[Ii]\.e|[Vv]s|[Cc]f)\.|\betc\.(?=\s+[a-z])/g;
+/** CommonMark fences: a backtick run with a backtick after it is inline code, and only a bare run as long closes. */
+function proseOnly(text) {
+    let fence = null;
+    let afterProse = false;
+    return text.split('\n').map(line => {
+        const trimmed = line.trimStart();
+        if (fence !== null) {
+            const close = trimmed.match(FENCE_CLOSE);
+            if (close && close[1][0] === fence[0] && close[1].length >= fence.length)
+                fence = null;
+            afterProse = false;
+            return '';
+        }
+        const open = trimmed.match(FENCE_OPEN);
+        if (open && !(open[1][0] === '`' && open[2].includes('`'))) {
+            fence = open[1];
+            afterProse = false;
+            return '';
+        }
+        // Quotes, table rows and indented code are not the speaker's own; indented code cannot interrupt a paragraph or list item.
+        const dropped = /^\s*[>|]/.test(line) || (!afterProse && /^(?: {4}|\t)(?!\s*(?:[-*]|\d+\.)\s)/.test(line));
+        afterProse = !dropped && trimmed !== '' && !/^#{1,6}(?:\s|$)/.test(trimmed);
+        return dropped ? '' : line;
+    }).join('\n');
 }
-function isLetterOrDigit(ch) {
-    return ch !== undefined && /[\p{L}\p{N}]/u.test(ch);
+function masked(text) {
+    return text.replace(NO_BREAK, span => 'x'.repeat(span.length));
 }
-/** The last plausible CLOSING single quote, or -1. Mirror of the opener rule in `boundToClause`: a closer sits tight
- * against its literal and is never followed by a letter, which is what makes "user's" an apostrophe. */
-function lastCloserIndex(full) {
-    // ONE pass, not one per apostrophe: a rescan per apostrophe goes quadratic
-    // on a transcript full of elisions ("keep 'em", "wait 'til", ...).
-    //
-    // A quote CLOSES THE SIDE IT IS TIGHT AGAINST, so the test needs both
-    // neighbours: these two have the same following character and opposite roles -
-    //   "preserve 'a, b'-style"   next '-'  -> CLOSER, content on the left
-    //   "then run '--force,"      next '-'  -> OPENER, content on the right
-    // so no forward-only rule can separate them.
-    //
-    //   never a closer   next is a letter/digit      "user's", "keep 'em"
-    //   closer           prev is non-whitespace      "'a, b' to", "'a, b'-style"
-    //   closer           next is whitespace or end   "preserve 'a, b ' exactly"
-    //   closer           next is clause punctuation  "preserve 'a, b ', then"
-    //                    that itself ends the token
-    //
-    // The last clause is what separates ", " from ".env": a dot followed by a
-    // letter joins a filename, a dot followed by space ends a sentence.
-    for (let j = full.length - 1; j >= 0; j--) {
-        if (full[j] !== "'")
-            continue;
-        const prev = full[j - 1];
-        const next = full[j + 1];
-        if (isLetterOrDigit(next))
-            continue;
-        // "tight against content" excludes an OPENING delimiter: in
-        // "call parse('--force" the paren is non-whitespace but the quote after
-        // it is an opener, and treating it as a closer let an earlier elision
-        // pair across the clause boundary.
-        const tightBefore = prev !== undefined && !/[\s([{]/.test(prev);
-        const endsAfter = next === undefined || /\s/.test(next);
-        // Punctuation splits in two:
-        //   , ; : ! ? ) ] }  never join tokens - they end the literal whatever
-        //                    follows, so "'echo a, b ';then" closes at the ;
-        //   . - _ / @ #      DO join tokens - "'.env" and "'--force" continue a
-        //                    filename or flag, so these only close when trailed
-        //                    by whitespace
-        const afterNext = full[j + 2];
-        const closesRegardless = next !== undefined && /[,;:!?)\]}]/.test(next);
-        const joinerThenSpace = next !== undefined && /[.\-_/@#]/.test(next) &&
-            (afterNext === undefined || /\s/.test(afterNext));
-        if (tightBefore || endsAfter || closesRegardless || joinerThenSpace)
-            return j;
-    }
-    return -1;
+/** A lowercase line continues the line above only when that line has not ended its sentence; "e.g." at a line end has not. */
+function continues(line, previous) {
+    if (previous === undefined || !/^[a-z]/.test(line) || LABEL.test(line))
+        return false;
+    return !SENTENCE_END.test(masked(`${previous} ${line}`).slice(0, previous.length));
 }
-/** Cuts `full` (keyword + content) at the first prose `,;:` or `[.!?]` followed by whitespace, scanning from `searchFrom` so a keyword's
- * own colon never counts; the whitespace rule keeps `.env` whole and `maxLen` truncates a long span the 500-char gate would drop. */
-function boundToClause(full, searchFrom, maxLen = 200) {
-    // Depth-awareness matters because memories are full of code; a naive scan cuts
-    //   "Always call build(x, y) before deploy."  ->  "Always call build(x"
-    // and the fragment PASSES the write gate, since code punctuation reads as "specific".
-    //
-    // So: a separator only ends the clause at bracket depth zero and outside
-    // quotes. Unbalanced closers are tolerated (depth floors at 0) because
-    // captured text often starts mid-expression.
-    const lastCloser = lastCloserIndex(full);
-    let depth = 0;
-    let quote = null;
-    let cutEnd = full.length;
-    for (let i = searchFrom; i < full.length; i++) {
-        const ch = full[i];
-        if (quote) {
-            if (closesQuote(full, i, quote))
-                quote = null;
-            continue;
-        }
-        if (ch === '"' || ch === '`') {
-            quote = ch;
-            continue;
-        }
-        if (ch === "'") {
-            if (opensSingleQuote(full, i, lastCloser)) {
-                quote = ch;
-            }
-            continue;
-        }
-        if (ch === '(' || ch === '[' || ch === '{') {
-            depth++;
-            continue;
-        }
-        if (ch === ')' || ch === ']' || ch === '}') {
-            if (depth > 0)
-                depth--;
-            continue;
-        }
-        if (depth > 0)
+function proseBlocks(text) {
+    const blocks = [];
+    let pending = [];
+    const flush = () => {
+        if (pending.length > 0)
+            blocks.push(pending.join(' '));
+        pending = [];
+    };
+    for (const line of text.split('\n')) {
+        const trimmed = line.trim();
+        if (!trimmed || /^#+\s/.test(trimmed)) {
+            flush();
             continue;
-        const next = full[i + 1];
-        // Prose clause separator: , ; : followed by whitespace.
-        if ((ch === ',' || ch === ';' || ch === ':') && next !== undefined && /\s/.test(next)) {
-            cutEnd = i;
-            break;
-        }
-        // Sentence terminator, but only when followed by whitespace or end — a
-        // bare [.!?] splits inside .env, capture.ts, v1.35.0.
-        if ((ch === '.' || ch === '!' || ch === '?') && (next === undefined || /\s/.test(next))) {
-            cutEnd = i + 1;
-            break;
         }
+        const bullet = trimmed.match(BULLET);
+        if (bullet || !continues(trimmed, pending.at(-1)))
+            flush();
+        pending.push(bullet?.[1] ?? trimmed);
     }
-    let bounded = full.slice(0, cutEnd);
-    if (bounded.length > maxLen)
-        bounded = bounded.slice(0, maxLen);
-    return bounded;
+    flush();
+    return blocks;
 }
-function closesQuote(full, i, quote) {
-    // Closing uses the SAME shape test as opening, else a possessive INSIDE a literal closes it early:
-    //   "Always pass 'user's a, b list' to the parser."  ->  "Always pass 'user's a"
-    return full[i] === quote && (quote !== "'" || !isLetterOrDigit(full[i + 1]));
+function completeSentences(text) {
+    // Masking keeps offsets, so each sentence is cut from the unchanged source.
+    const segments = new Intl.Segmenter('en', { granularity: 'sentence' }).segment(masked(text));
+    return [...segments].map(({ index, segment }) => text.slice(index, index + segment.length).trim());
 }
-function opensSingleQuote(full, i, lastCloser) {
-    // Quote handling has to tell an APOSTROPHE from a single-quoted
-    // LITERAL, because getting either wrong reintroduces the fragment
-    // defect this change exists to remove, and both fragments PASS the
-    // write gate (code punctuation reads as "specific"):
-    //   treat every ' as a quote  -> "Always ensure it's enabled, then
-    //     restart..." never leaves quote mode, bounding disabled entirely
-    //   treat no ' as a quote     -> "Always pass 'a, b' to the parser."
-    //     cuts at the comma and stores "Always pass 'a"
-    //
-    // Discriminator: a ' OPENS a literal only at a word boundary - preceded
-    // by start/whitespace/open-bracket AND followed by non-whitespace. An
-    // in-word apostrophe ("it's", "user's") has letters on both sides and is
-    // just a character.
-    const prev = i > 0 ? full[i - 1] : undefined;
-    const next = full[i + 1];
-    const atBoundary = (prev === undefined || /[\s([{]/.test(prev)) &&
-        next !== undefined && !/\s/.test(next);
-    // Pairing is VERIFIED, not assumed. A word-boundary test alone still
-    // opens quote mode on elided forms ("keep 'em", "wait 'til"), which
-    // have no closer, so the scanner never leaves quote mode and bounding
-    // is disabled for the rest of the capture. Requiring an actual closing
-    // quote later in the string replaces a guess with a checkable fact -
-    // an elided form simply has no partner.
-    // ...and the partner must LOOK like a closer, not merely be another
-    // apostrophe. Distance cannot separate the two cases - both put a `'`
-    // far downstream:
-    //   "Always pass '<600-char literal>' to the parser"  -> the far ' IS
-    //     the closer, and pairing must succeed
-    //   "keep 'em enabled, then <520 chars> check user's config" -> the far
-    //     ' is in-word, and pairing against it re-opens quote mode on the
-    //     elision, disabling bounding for the rest of the capture
-    // So apply the SAME shape rule already used to open a literal, mirrored:
-    // a closer has non-whitespace before it and whitespace/punctuation (not
-    // a letter) after. "user's" fails it on both counts.
-    return atBoundary && lastCloser > i;
+function insideBrackets(lead) {
+    return (lead.match(/[([{]/g) ?? []).length > (lead.match(/[)\]}]/g) ?? []).length;
 }
-function cleanExtract(raw) {
-    let content = raw
-        .replace(/^[:\s-]+/, '')
-        .replace(/[.!?,;:\s]+$/, '')
-        .trim();
-    // Trailing cleanup: clause-bounding can cut inside a parenthetical and
-    // leave an unmatched trailing ')' (e.g. "...(two had never got entries),"
-    // bounds to "never got entries)"). Strip a trailing ')' ONLY while closes
-    // outnumber opens in the string so far — a balanced parenthetical like
-    // "always run the suite (twice)" must be left intact.
-    while (content.endsWith(')')) {
-        const opens = (content.match(/\(/g) ?? []).length;
-        const closes = (content.match(/\)/g) ?? []).length;
-        if (closes <= opens)
-            break;
-        content = content
-            .slice(0, -1)
-            .replace(/[.!?,;:\s]+$/, '')
-            .trim();
-    }
-    return content;
-}
-/** The 500-char write gate minus the 200-char bound counted from the keyword. */
-const RULE_LEAD_CAP = 300;
-/** A modal rule keeps its subject ("We must never ..."); null keeps the keyword-start form when the lead is long or the keyword sits inside a bracket. */
-function ruleLead(sentence, keywordStart, contentStart) {
-    if (keywordStart < 0 || contentStart < 0 || contentStart > RULE_LEAD_CAP)
+function extractFromPatterns(sentence, patterns, category) {
+    // A keyword inside brackets does not support a detached assertion; a later one outside them still does.
+    const keyed = patterns.some((pattern) => [...sentence.matchAll(new RegExp(pattern.source, `${pattern.flags}g`))]
+        .some((match) => !insideBrackets(sentence.slice(0, match.index))));
+    if (!keyed)
         return null;
-    let depth = 0;
-    for (const ch of sentence.slice(0, keywordStart)) {
-        if (ch === '(' || ch === '[' || ch === '{')
-            depth++;
-        else if ((ch === ')' || ch === ']' || ch === '}') && depth > 0)
-            depth--;
-    }
-    return depth > 0 ? null : sentence.slice(0, contentStart);
-}
-function extractFromPatterns(sentence, patterns, category, tag) {
-    for (const pat of patterns) {
-        // `d` gives per-group offsets; see the contentStart comment below.
-        const dpat = pat.flags.includes('d') ? pat : new RegExp(pat.source, pat.flags + 'd');
-        const match = dpat.exec(sentence);
-        if (match) {
-            let bounded;
-            if (pat === PREFERENCE_PATTERNS[0]) {
-                // PREFERENCE_PATTERNS[0] skips clause-bounding (see comment at its definition).
-                bounded = match[1] ?? match[0];
-            }
-            else {
-                // group1 = keyword + separator (verbatim, never clause-bounded);
-                // group2 = the content that follows it (clause-bounded below).
-                //
-                // The keyword is kept only when it carries SEMANTIC SIGN, a
-                // negation or modality ("never", "must not", "do not ever",
-                // "always"), because dropping those inverts the meaning. A LABEL keyword ("decision:", "rule:", "error:",
-                // "important:") carries no sign: it only names the category, which
-                // is already recorded in `category`/`tags`, so prefixing it onto the
-                // content is duplication that also breaks value-keyed matching
-                // (the rejected-value digest hashes the bare content).
-                // Discriminator: a label ends in its own colon, or is a "the X is"
-                // phrase.
-                const rawPrefix = match[1] ?? '';
-                // ONLY colon-terminated labels are dropped: dropping "the X is/was" leaves a residue starting
-                // with "to ", which isFragment rejects, so "The plan is to ship on Friday" would store nothing.
-                const isLabelPrefix = /:\s*$/.test(rawPrefix);
-                const keywordPrefix = isLabelPrefix ? '' : rawPrefix;
-                // Scan the UNTRUNCATED remainder, not match[2]: a quoted literal whose closer sits past the
-                // 500-char group cap would read as prose; boundToClause caps its OUTPUT, so this costs nothing.
-                // Group 2's REAL offset from the regex engine: DECISION_PATTERNS carry an uncaptured subject
-                // ("we ", "let's ") ahead of group 1, so `match.index + rawPrefix.length` lands inside the keyword.
-                const contentStart = match.indices?.[2]?.[0] ?? -1;
-                const afterKeyword = contentStart >= 0 ? sentence.slice(contentStart) : (match[2] ?? match[0]);
-                const keywordStart = match.indices?.[1]?.[0] ?? -1;
-                const lead = pat === RULE_PATTERNS[0] ? ruleLead(sentence, keywordStart, contentStart) : null;
-                bounded = lead !== null
-                    ? boundToClause(lead + afterKeyword, lead.length, keywordStart + 200)
-                    : boundToClause(keywordPrefix + afterKeyword, keywordPrefix.length);
-            }
-            const content = cleanExtract(bounded);
-            if (content.length >= 8 && content.length <= 500) {
-                return { content, category, tags: [tag, 'captured'] };
-            }
-        }
-    }
-    return null;
+    const content = sentence
+        .replace(/^(?:decision|plan|rule|important|critical|remember|error|bug|gotcha|watch out|careful|warning|caveat|trap|don't forget|easy to miss):\s*/i, '')
+        .replace(/[.!?\s]+$/, '').trim();
+    return content.length <= MAX_CHARS ? { content, category, tags: [category, 'captured'] } : null;
 }
-/** Extract spec items from bullet lists that follow spec-like headings. */
 function extractSpecSections(text) {
-    const items = [];
-    const lines = text.split('\n');
+    const bullets = [];
     let inSpecSection = false;
-    for (const line of lines) {
+    let open = false;
+    for (const line of text.split('\n')) {
         const trimmed = line.trim();
-        // Check if this line is a spec heading
-        if (SPEC_HEADING_PATTERNS.some((p) => p.test(trimmed))) {
+        if (SPEC_HEADING_PATTERNS.some(pattern => pattern.test(trimmed))) {
             inSpecSection = true;
+            open = false;
             continue;
         }
-        // Another heading resets the section
-        if (/^#+\s/.test(trimmed) || /^[A-Z][a-z]+:$/.test(trimmed)) {
+        if (/^#+\s/.test(trimmed) || /^[A-Z][a-z]+:$/.test(trimmed))
             inSpecSection = false;
-            continue;
-        }
-        // Blank line after non-bullet content ends section
-        if (!trimmed && inSpecSection) {
-            // Keep going, blank lines within spec sections are ok
-            continue;
-        }
-        if (inSpecSection) {
-            const bulletMatch = trimmed.match(/^[-*]\s+(.+)/) || trimmed.match(/^\d+\.\s+(.+)/);
-            if (bulletMatch) {
-                const content = bulletMatch[1].trim();
-                if (content.length >= 8 && content.length <= 500) {
-                    items.push({
-                        content,
-                        category: 'spec',
-                        tags: ['spec', 'captured'],
-                    });
+        const bullet = trimmed.match(BULLET);
+        const wrapped = inSpecSection && open && !bullet && trimmed !== '' && (/^\s/.test(line) || continues(trimmed, bullets.at(-1)));
+        if (inSpecSection && bullet)
+            bullets.push(bullet[1].trim());
+        else if (wrapped)
+            bullets[bullets.length - 1] += ` ${trimmed}`;
+        open = inSpecSection && (bullet !== null || wrapped);
+    }
+    return bullets
+        .filter(content => content.length <= MAX_CHARS && !content.endsWith('?'))
+        .map(content => ({ content, category: 'spec', tags: ['spec', 'captured'] }));
+}
+function extractOne(text) {
+    const prose = proseOnly(text);
+    const items = extractSpecSections(prose);
+    for (const block of proseBlocks(prose)) {
+        for (const sentence of completeSentences(block)) {
+            // A question asks rather than states; far over the bound, skip before any pattern runs.
+            if (sentence.endsWith('?') || sentence.length > 2 * MAX_CHARS)
+                continue;
+            for (const [category, patterns] of CATEGORIES) {
+                const item = extractFromPatterns(sentence, patterns, category);
+                if (item) {
+                    items.push(item);
+                    break;
                 }
             }
         }
     }
     return items;
 }
-/**
- * Main extraction function. Scans text for actionable items using heuristics.
- */
-export function extractFromText(text) {
-    const items = [];
+/** Captures complete supported statements, each text parsed alone so a fence or heading in one never reaches the next. */
+export function extractFromTexts(texts) {
     const seen = new Set();
-    const addIfNew = (item) => {
-        const norm = duplicateKey(item.content);
-        if (seen.has(norm))
-            return;
-        if (!isContentWorthStoring(item.content))
-            return;
-        seen.add(norm);
-        items.push(item);
-    };
-    // 1. Extract spec sections (bullet lists under spec headings)
-    for (const item of extractSpecSections(text)) {
-        addIfNew(item);
-    }
-    // 2. Pattern-match on individual sentences
-    const sentences = splitSentences(text);
-    for (const sentence of sentences) {
-        // Try each category in priority order
-        const decision = extractFromPatterns(sentence, DECISION_PATTERNS, 'decision', 'decision');
-        if (decision) {
-            addIfNew(decision);
-            continue;
-        }
-        const rule = extractFromPatterns(sentence, RULE_PATTERNS, 'rule', 'rule');
-        if (rule) {
-            addIfNew(rule);
-            continue;
-        }
-        const error = extractFromPatterns(sentence, ERROR_PATTERNS, 'error', 'error');
-        if (error) {
-            addIfNew(error);
-            continue;
-        }
-        const preference = extractFromPatterns(sentence, PREFERENCE_PATTERNS, 'preference', 'preference');
-        if (preference) {
-            addIfNew(preference);
-            continue;
-        }
-    }
-    return items;
+    return texts.flatMap(extractOne).filter((item) => {
+        const key = duplicateKey(item.content);
+        if (seen.has(key) || !assessAutomaticMemory(item.content).accepted)
+            return false;
+        seen.add(key);
+        return true;
+    });
+}
+export function extractFromText(text) {
+    return extractFromTexts([text]);
 }
 //# sourceMappingURL=extract.js.map

package/dist/capture/failure-caller.d.ts

@@ -0,0 +1,21 @@
+import type { Context } from '../api/types.js';
+import type { CallerProject } from '../prompt-hook.js';
+import { type CaptureErrorOutcome, type RoutineRule } from './failure-reading.js';
+type FailureSkip = Exclude<CaptureErrorOutcome, 'stored' | 'duplicate'>;
+/** failureReport's reading of the payload, read on the caller's machine, plus the id its retries carry. */
+export interface CallerFailureRequest {
+    readonly sessionId: string;
+    readonly project: CallerProject;
+    readonly tool: string | null;
+    readonly text: string | null;
+    readonly skip: FailureSkip | null;
+    readonly rule: RoutineRule | null;
+    readonly detailHash: string | null;
+    readonly requestId: string;
+}
+export interface CallerFailureResult {
+    readonly outcome: CaptureErrorOutcome;
+}
+export declare function captureFailureForCaller(ctx: Context, req: CallerFailureRequest): CallerFailureResult;
+export {};
+//# sourceMappingURL=failure-caller.d.ts.map

package/dist/capture/failure-caller.js

@@ -0,0 +1,74 @@
+// PostToolUseFailure for a caller on another machine: the lesson and its log row carry the caller's owner and project, and a retried request id gets the first answer.
+import { BadRequestError } from '../api-errors.js';
+import { storeLesson } from '../capture-error.js';
+import { recordFailure, requestOutcome, settleFailureOutcome } from '../failure-log.js';
+import { errorMessage, log } from '../log.js';
+import { scrubForSharing } from '../share-scrub.js';
+import { truncateCodePointSafe } from '../transcript-tail.js';
+import { assertRequestId, bindCaller, withCallerDb } from './caller-session.js';
+import { FAILURE_TEXT_MAX_CHARS, failureHash } from './failure-reading.js';
+const SKIPS = { 'skipped-interrupt': true, 'skipped-routine': true, 'skipped-invalid': true };
+const ROUTINE_RULES = { declined: true, 'os-permission': true, 'no-match': true, 'search-tool': true, 'quiet-exit': true };
+/** The shapes lessonFromFailure gives: text for a lesson or a routine skip, none for an interrupt or an unreadable payload, a rule only on a routine skip. */
+function checkedFailure(req) {
+    assertRequestId(req.requestId);
+    if (req.detailHash !== null && !/^[0-9a-f]{16}$/.test(req.detailHash))
+        throw new BadRequestError('detail hash: 16 lowercase hex characters or null');
+    const { skip, text, rule } = req;
+    if (skip !== null && !Object.hasOwn(SKIPS, skip))
+        throw new BadRequestError('skip: skipped-interrupt, skipped-routine, skipped-invalid or null');
+    if ((skip === 'skipped-routine') !== (rule !== null) || (rule !== null && !Object.hasOwn(ROUTINE_RULES, rule))) {
+        throw new BadRequestError('rule: a routine rule on skipped-routine, else null');
+    }
+    if (text === null || text.trim() === '') {
+        if (skip === null || skip === 'skipped-routine')
+            throw new BadRequestError('text: required for a lesson or a routine skip');
+        return { skip, text: null };
+    }
+    if (skip === 'skipped-interrupt' || skip === 'skipped-invalid')
+        throw new BadRequestError('text: null for an interrupt or an unreadable failure');
+    if (text.length > FAILURE_TEXT_MAX_CHARS)
+        throw new BadRequestError(`text: at most ${FAILURE_TEXT_MAX_CHARS} characters`);
+    // Scrubbed again, as the caller's scrub is not trusted, and cut after it, as a mask can be longer than what it hides.
+    return { skip, text: truncateCodePointSafe(scrubForSharing(text), FAILURE_TEXT_MAX_CHARS) };
+}
+export function captureFailureForCaller(ctx, req) {
+    const failure = checkedFailure(req);
+    const key = bindCaller(ctx, req.sessionId, req.project);
+    const earlier = withCallerDb(ctx, (db) => requestOutcome(db, ctx.tenantId, req.requestId, req.sessionId));
+    // A retry after a lost reply gets the first answer and writes nothing; one whose store failed tries the store again.
+    if (earlier !== null && earlier !== 'store-failed')
+        return { outcome: earlier };
+    let logged = 'store-failed';
+    try {
+        const caller = { actor: key.owner, originProject: req.project.name, origins: key.project };
+        const outcome = failure.skip === null ? storeLesson(ctx.hippoRoot, ctx.tenantId, failure.text, caller) : failure.skip;
+        logged = outcome;
+        return { outcome };
+    }
+    finally {
+        // The hash is of the text as stored, so a repeat counts against the lesson it repeats.
+        logOutcome(ctx, { ...req, text: failure.text }, key, logged, earlier !== null);
+    }
+}
+/** Runs after the outcome in its own try, so a log error never hides the store's error or its outcome. */
+function logOutcome(ctx, req, key, outcome, retried) {
+    try {
+        withCallerDb(ctx, (db) => {
+            // The request id allows one row, so a retry rewrites the first try's outcome.
+            if (retried) {
+                settleFailureOutcome(db, ctx.tenantId, req.requestId, outcome);
+                return;
+            }
+            recordFailure(db, {
+                tenantId: ctx.tenantId, sessionId: req.sessionId, tool: req.tool, outcome, rule: req.rule,
+                sigHash: req.text?.trim() ? failureHash(req.text) : null, detailHash: req.detailHash,
+                ownerSubject: key.owner, originProject: req.project.name, requestId: req.requestId,
+            });
+        });
+    }
+    catch (err) {
+        log.warn(`capture-error: failure not logged: ${errorMessage(err)}`);
+    }
+}
+//# sourceMappingURL=failure-caller.js.map