todoosy 0.3.4 → 0.3.7

package/dist/parser.js

@@ -1,11 +1,9 @@
 /**
  * Todoosy Parser
  */
+import { resolveNow, resolveKeyword, tryParseRelative } from './relative-date.js';
 const HEADING_REGEX = /^(#{1,6})\s+(.*)$/;
 const LIST_ITEM_REGEX = /^(\s*)([-*]|\d+\.)\s+(.*)$/;
-const DUE_ISO_REGEX = /^due\s+(\d{4})-(\d{2})-(\d{2})$/i;
-const DUE_US_REGEX = /^due\s+(\d{1,2})\/(\d{1,2})\/(\d{4})$/i;
-const DUE_US_SHORT_REGEX = /^due\s+(\d{1,2})\/(\d{1,2})\/(\d{2})$/i;
 const PRIORITY_REGEX = /^p(\d+)$/i;
 const ESTIMATE_REGEX = /^(\d+)([mhd])$/i;
 const HASHTAG_REGEX = /^#([a-zA-Z][a-zA-Z0-9_-]*)$/;
@@ -25,33 +23,24 @@ const MONTH_NAMES = {
 };
 // Built-in progress states (normalized to lowercase)
 const PROGRESS_STATES = new Set(['done', 'deleted', 'in progress', 'blocked']);
-function inferYear(month, day) {
-    const now = new Date();
-    const currentYear = now.getFullYear();
-    const currentMonth = now.getMonth() + 1;
-    const currentDay = now.getDate();
-    // Calculate months difference
-    let monthsDiff = (currentMonth - month) + (currentDay > day ? 0 : 0);
-    if (currentMonth > month || (currentMonth === month && currentDay > day)) {
-        monthsDiff = (currentMonth - month) + (currentDay > day ? 0 : -1);
-        // More precise: is the date more than 3 months in the past?
-        const candidateDate = new Date(currentYear, month - 1, day);
-        const threeMonthsAgo = new Date(now);
-        threeMonthsAgo.setMonth(threeMonthsAgo.getMonth() - 3);
-        if (candidateDate < threeMonthsAgo) {
-            return currentYear + 1;
-        }
+function inferYear(month, day, now) {
+    const [todayY, todayM, todayD] = now.todayIso.split('-').map(n => parseInt(n, 10));
+    const candidate = new Date(Date.UTC(todayY, month - 1, day));
+    // Three months ago in UTC
+    const threeMonthsAgo = new Date(Date.UTC(todayY, todayM - 1 - 3, todayD));
+    if (candidate < threeMonthsAgo) {
+        return todayY + 1;
     }
-    return currentYear;
+    return todayY;
 }
-function parseDate(dateStr) {
+function parseISODate(dateStr) {
     // ISO format: YYYY-MM-DD or YYYY-M-D
     const isoMatch = dateStr.match(/^(\d{4})-(\d{1,2})-(\d{1,2})$/);
     if (isoMatch) {
         const year = isoMatch[1];
         const month = isoMatch[2].padStart(2, '0');
         const day = isoMatch[3].padStart(2, '0');
-        return { date: `${year}-${month}-${day}`, valid: true, raw: dateStr };
+        return `${year}-${month}-${day}`;
     }
     // Short ISO format: YY-MM-DD or YY-M-D
     const isoShortMatch = dateStr.match(/^(\d{2})-(\d{1,2})-(\d{1,2})$/);
@@ -59,7 +48,7 @@ function parseDate(dateStr) {
         const year = `20${isoShortMatch[1]}`;
         const month = isoShortMatch[2].padStart(2, '0');
         const day = isoShortMatch[3].padStart(2, '0');
-        return { date: `${year}-${month}-${day}`, valid: true, raw: dateStr };
+        return `${year}-${month}-${day}`;
     }
     // Year-first with slashes: YYYY/MM/DD
     const ymdSlashMatch = dateStr.match(/^(\d{4})\/(\d{1,2})\/(\d{1,2})$/);
@@ -67,10 +56,9 @@ function parseDate(dateStr) {
         const year = ymdSlashMatch[1];
         const month = ymdSlashMatch[2].padStart(2, '0');
         const day = ymdSlashMatch[3].padStart(2, '0');
-        return { date: `${year}-${month}-${day}`, valid: true, raw: dateStr };
+        return `${year}-${month}-${day}`;
     }
-    // Slash format: X/X/X - need smart heuristics to determine format
-    // Could be YY/MM/DD, MM/DD/YY, DD/MM/YY, MM/DD/YYYY, DD/MM/YYYY
+    // Slash format with heuristics
     const slashMatch = dateStr.match(/^(\d{1,4})\/(\d{1,2})\/(\d{1,4})$/);
     if (slashMatch) {
         const first = parseInt(slashMatch[1], 10);
@@ -79,16 +67,13 @@ function parseDate(dateStr) {
         const firstLen = slashMatch[1].length;
         const thirdLen = slashMatch[3].length;
         let year, month, day;
-        // 4-digit year at start: YYYY/MM/DD
         if (firstLen === 4) {
             year = first;
             month = second;
             day = third;
         }
-        // 4-digit year at end: XX/XX/YYYY
         else if (thirdLen === 4) {
             year = third;
-            // Heuristic: if first > 12, must be DD/MM/YYYY, otherwise MM/DD/YYYY
             if (first > 12) {
                 day = first;
                 month = second;
@@ -98,36 +83,25 @@ function parseDate(dateStr) {
                 day = second;
             }
         }
-        // 2-digit year - need to determine position
         else if (firstLen === 2 && thirdLen === 2) {
-            // Smart heuristics:
-            // - If first > 31, must be YY/MM/DD (year at start)
-            // - If first > 12, must be DD/MM/YY (day at start)
-            // - Otherwise treat as MM/DD/YY (US convention)
             if (first > 31) {
-                // YY/MM/DD
                 year = 2000 + first;
                 month = second;
                 day = third;
             }
             else if (first > 12) {
-                // DD/MM/YY
                 day = first;
                 month = second;
                 year = 2000 + third;
             }
             else {
-                // MM/DD/YY (US convention)
                 month = first;
                 day = second;
                 year = 2000 + third;
             }
         }
-        // Single digit somewhere
         else {
-            // Default to MM/DD/YY if year at end seems 2-digit, else YYYY/MM/DD
             if (thirdLen <= 2) {
-                // XX/XX/YY
                 if (first > 12) {
                     day = first;
                     month = second;
@@ -139,178 +113,243 @@ function parseDate(dateStr) {
                 year = 2000 + third;
             }
             else {
-                // Shouldn't happen, but handle it
                 year = first;
                 month = second;
                 day = third;
             }
         }
-        return {
-            date: `${year}-${String(month).padStart(2, '0')}-${String(day).padStart(2, '0')}`,
-            valid: true,
-            raw: dateStr
-        };
+        return `${year}-${String(month).padStart(2, '0')}-${String(day).padStart(2, '0')}`;
     }
-    return { date: null, valid: false, raw: dateStr };
+    return null;
 }
-function parseTextDate(parts) {
-    if (parts.length < 2) {
-        return { date: null, valid: false, partsConsumed: 0 };
-    }
-    // Try "Month Day [Year]" format first
-    const monthStr = parts[0].toLowerCase();
-    let month = MONTH_NAMES[monthStr];
-    if (month !== undefined) {
-        const dayStr = parts[1];
-        const dayMatch = dayStr.match(/^(\d{1,2})$/);
+function parseTextDate(parts, startIdx, now) {
+    if (startIdx + 1 >= parts.length)
+        return null;
+    // "Month Day [Year]"
+    const monthStr = parts[startIdx].toLowerCase();
+    const monthFromName = MONTH_NAMES[monthStr];
+    if (monthFromName !== undefined) {
+        const dayMatch = parts[startIdx + 1].match(/^(\d{1,2})$/);
         if (dayMatch) {
             const day = parseInt(dayMatch[1], 10);
             if (day >= 1 && day <= 31) {
-                // Check for year (4-digit or 2-digit)
-                if (parts.length >= 3) {
-                    const yearStr = parts[2];
-                    const yearMatch = yearStr.match(/^(\d{4})$/);
-                    if (yearMatch) {
-                        const year = parseInt(yearMatch[1], 10);
-                        const monthPadded = String(month).padStart(2, '0');
-                        const dayPadded = String(day).padStart(2, '0');
-                        return { date: `${year}-${monthPadded}-${dayPadded}`, valid: true, partsConsumed: 3 };
+                if (startIdx + 2 < parts.length) {
+                    const yearStr = parts[startIdx + 2];
+                    const yMatch4 = yearStr.match(/^(\d{4})$/);
+                    if (yMatch4) {
+                        return { date: `${yMatch4[1]}-${String(monthFromName).padStart(2, '0')}-${String(day).padStart(2, '0')}`, partsConsumed: 3 };
                     }
-                    // Try 2-digit year
-                    const yearShortMatch = yearStr.match(/^(\d{2})$/);
-                    if (yearShortMatch) {
-                        const year = 2000 + parseInt(yearShortMatch[1], 10);
-                        const monthPadded = String(month).padStart(2, '0');
-                        const dayPadded = String(day).padStart(2, '0');
-                        return { date: `${year}-${monthPadded}-${dayPadded}`, valid: true, partsConsumed: 3 };
+                    const yMatch2 = yearStr.match(/^(\d{2})$/);
+                    if (yMatch2) {
+                        return { date: `${2000 + parseInt(yMatch2[1], 10)}-${String(monthFromName).padStart(2, '0')}-${String(day).padStart(2, '0')}`, partsConsumed: 3 };
                     }
                 }
-                // No year provided, infer it
-                const year = inferYear(month, day);
-                const monthPadded = String(month).padStart(2, '0');
-                const dayPadded = String(day).padStart(2, '0');
-                return { date: `${year}-${monthPadded}-${dayPadded}`, valid: true, partsConsumed: 2 };
+                const year = inferYear(monthFromName, day, now);
+                return { date: `${year}-${String(monthFromName).padStart(2, '0')}-${String(day).padStart(2, '0')}`, partsConsumed: 2 };
             }
         }
     }
-    // Try "Day Month [Year]" format
-    const dayFirstMatch = parts[0].match(/^(\d{1,2})$/);
+    // "Day Month [Year]"
+    const dayFirstMatch = parts[startIdx].match(/^(\d{1,2})$/);
     if (dayFirstMatch) {
         const day = parseInt(dayFirstMatch[1], 10);
         if (day >= 1 && day <= 31) {
-            const monthStr2 = parts[1].toLowerCase();
-            month = MONTH_NAMES[monthStr2];
-            if (month !== undefined) {
-                // Check for year (4-digit or 2-digit)
-                if (parts.length >= 3) {
-                    const yearStr = parts[2];
-                    const yearMatch = yearStr.match(/^(\d{4})$/);
-                    if (yearMatch) {
-                        const year = parseInt(yearMatch[1], 10);
-                        const monthPadded = String(month).padStart(2, '0');
-                        const dayPadded = String(day).padStart(2, '0');
-                        return { date: `${year}-${monthPadded}-${dayPadded}`, valid: true, partsConsumed: 3 };
+            const monthStr2 = parts[startIdx + 1].toLowerCase();
+            const monthFromName2 = MONTH_NAMES[monthStr2];
+            if (monthFromName2 !== undefined) {
+                if (startIdx + 2 < parts.length) {
+                    const yearStr = parts[startIdx + 2];
+                    const yMatch4 = yearStr.match(/^(\d{4})$/);
+                    if (yMatch4) {
+                        return { date: `${yMatch4[1]}-${String(monthFromName2).padStart(2, '0')}-${String(day).padStart(2, '0')}`, partsConsumed: 3 };
                     }
-                    // Try 2-digit year
-                    const yearShortMatch = yearStr.match(/^(\d{2})$/);
-                    if (yearShortMatch) {
-                        const year = 2000 + parseInt(yearShortMatch[1], 10);
-                        const monthPadded = String(month).padStart(2, '0');
-                        const dayPadded = String(day).padStart(2, '0');
-                        return { date: `${year}-${monthPadded}-${dayPadded}`, valid: true, partsConsumed: 3 };
+                    const yMatch2 = yearStr.match(/^(\d{2})$/);
+                    if (yMatch2) {
+                        return { date: `${2000 + parseInt(yMatch2[1], 10)}-${String(monthFromName2).padStart(2, '0')}-${String(day).padStart(2, '0')}`, partsConsumed: 3 };
                     }
                 }
-                // No year provided, infer it
-                const year = inferYear(month, day);
-                const monthPadded = String(month).padStart(2, '0');
-                const dayPadded = String(day).padStart(2, '0');
-                return { date: `${year}-${monthPadded}-${dayPadded}`, valid: true, partsConsumed: 2 };
+                const year = inferYear(monthFromName2, day, now);
+                return { date: `${year}-${String(monthFromName2).padStart(2, '0')}-${String(day).padStart(2, '0')}`, partsConsumed: 2 };
             }
         }
     }
-    return { date: null, valid: false, partsConsumed: 0 };
+    return null;
 }
-export function parseTokensInParenGroup(content, groupStart) {
+/**
+ * Try to parse a single date (keyword | ISO/slash | text date | relative offset)
+ * starting at parts[startIdx]. Returns the date and parts consumed, or null.
+ */
+function tryParseSingleDate(parts, startIdx, now) {
+    if (startIdx >= parts.length)
+        return null;
+    // Keywords
+    const kw = resolveKeyword(parts[startIdx], now);
+    if (kw !== null)
+        return { date: kw, partsConsumed: 1 };
+    // ISO/slash single-token
+    const iso = parseISODate(parts[startIdx]);
+    if (iso !== null)
+        return { date: iso, partsConsumed: 1 };
+    // Text date (multi-token)
+    const textDate = parseTextDate(parts, startIdx, now);
+    if (textDate !== null)
+        return textDate;
+    // Relative offset
+    const rel = tryParseRelative(parts, startIdx, now);
+    if (rel !== null)
+        return rel;
+    return null;
+}
+/**
+ * Pre-tokenize parens content. The base split is by commas/whitespace; then
+ * each part is further split on `~` so that `~` always appears as its own
+ * delimiter token. Examples:
+ *   `~Apr`              → [`~`, `Apr`]
+ *   `Apr 24~Apr 27`     → [`Apr`, `24`, `~`, `Apr`, `27`]  (input parts: [`Apr`, `24~Apr`, `27`])
+ *   `2026-04-24~2026-04-27` → [`2026-04-24`, `~`, `2026-04-27`]
+ */
+function tokenizeContent(content) {
+    const parts = [];
+    const partOffsets = [];
+    let i = 0;
+    while (i < content.length) {
+        // Skip separators (commas + whitespace)
+        while (i < content.length && (content[i] === ',' || /\s/.test(content[i])))
+            i++;
+        if (i >= content.length)
+            break;
+        // Read until next separator
+        const tokenStart = i;
+        while (i < content.length && content[i] !== ',' && !/\s/.test(content[i]))
+            i++;
+        const raw = content.slice(tokenStart, i);
+        // Split this token on `~`, emitting `~` as its own part. Preserve offsets.
+        let cursor = 0;
+        while (cursor < raw.length) {
+            const tildeIdx = raw.indexOf('~', cursor);
+            if (tildeIdx === -1) {
+                const seg = raw.slice(cursor);
+                if (seg.length > 0) {
+                    parts.push(seg);
+                    partOffsets.push(tokenStart + cursor);
+                }
+                break;
+            }
+            // Pre-tilde segment (may be empty)
+            if (tildeIdx > cursor) {
+                parts.push(raw.slice(cursor, tildeIdx));
+                partOffsets.push(tokenStart + cursor);
+            }
+            // The tilde itself
+            parts.push('~');
+            partOffsets.push(tokenStart + tildeIdx);
+            cursor = tildeIdx + 1;
+        }
+    }
+    return { parts, partOffsets };
+}
+export function parseTokensInParenGroup(content, groupStart, now) {
+    if (!now)
+        now = resolveNow(undefined, undefined);
+    return parseTokensInParenGroupImpl(content, groupStart, now);
+}
+function parseTokensInParenGroupImpl(content, groupStart, now) {
     const tokens = [];
-    // Split by comma and/or whitespace
-    const parts = content.split(/[,\s]+/).filter(p => p.length > 0);
-    let currentPos = 0;
+    const { parts, partOffsets } = tokenizeContent(content);
+    // Compute the absolute end of part i.
+    const partEndOffset = (i) => partOffsets[i] + parts[i].length;
+    const absStart = (i) => groupStart + 1 + partOffsets[i];
+    const absEnd = (i) => groupStart + 1 + partEndOffset(i);
     const skipIndices = new Set();
     for (let i = 0; i < parts.length; i++) {
         if (skipIndices.has(i))
             continue;
         const part = parts[i];
-        const partStart = content.indexOf(part, currentPos);
-        const absoluteStart = groupStart + 1 + partStart; // +1 for opening paren
-        const absoluteEnd = absoluteStart + part.length;
-        currentPos = partStart + part.length;
-        // Check for due date
+        // Standalone `~` here means we encountered a tilde without a leading date —
+        // either a soft-prefix introducer (`~<date>`) or an orphan. Try as soft prefix.
+        if (part === '~') {
+            // Soft prefix: try to consume a single date (no span — span would have a leading date).
+            const inner = tryParseSingleDate(parts, i + 1, now);
+            if (inner !== null) {
+                const startIndex = i;
+                const endIndex = i + inner.partsConsumed; // last consumed index = endIndex
+                // Mark consumed
+                for (let j = i + 1; j <= endIndex; j++)
+                    skipIndices.add(j);
+                const raw = content.slice(partOffsets[startIndex], partEndOffset(endIndex));
+                tokens.push({
+                    type: 'due',
+                    value: inner.date,
+                    raw,
+                    start: absStart(startIndex),
+                    end: absEnd(endIndex),
+                    soft: true,
+                });
+                continue;
+            }
+            // Orphan `~` — drop silently.
+            continue;
+        }
+        // 'due' keyword
         if (part.toLowerCase() === 'due') {
-            // Look for the next part(s) as the date
-            const remainingParts = parts.slice(i + 1);
-            if (remainingParts.length > 0) {
-                // Check for soft date prefix (~)
-                let isSoft = false;
-                let datePartsToCheck = remainingParts;
-                if (remainingParts[0].startsWith('~')) {
-                    isSoft = true;
-                    // Remove the tilde for parsing
-                    datePartsToCheck = [remainingParts[0].slice(1), ...remainingParts.slice(1)];
-                }
-                // First try standard date formats (single part)
-                const dateResult = parseDate(datePartsToCheck[0]);
-                if (dateResult.valid) {
-                    const nextPartStart = content.indexOf(remainingParts[0], currentPos);
-                    const nextAbsoluteEnd = groupStart + 1 + nextPartStart + remainingParts[0].length;
-                    tokens.push({
-                        type: 'due',
-                        value: dateResult.date,
-                        raw: `due ${remainingParts[0]}`,
-                        start: absoluteStart,
-                        end: nextAbsoluteEnd,
-                        soft: isSoft || undefined,
-                    });
-                    skipIndices.add(i + 1);
-                    continue;
-                }
-                // Try text date formats (multiple parts: Month Day [Year])
-                const textDateResult = parseTextDate(datePartsToCheck);
-                if (textDateResult.valid) {
-                    // Calculate the end position
-                    let rawParts = [`due`];
-                    let endPos = currentPos;
-                    for (let j = 0; j < textDateResult.partsConsumed; j++) {
-                        rawParts.push(remainingParts[j]);
-                        skipIndices.add(i + 1 + j);
-                        endPos = content.indexOf(remainingParts[j], endPos) + remainingParts[j].length;
+            // Look ahead. After `due` we may have `~<date>`, `<date>~<date>`, or `<date>`.
+            let cursor = i + 1;
+            let isSoftPrefix = false;
+            if (cursor < parts.length && parts[cursor] === '~') {
+                isSoftPrefix = true;
+                cursor++;
+            }
+            const first = tryParseSingleDate(parts, cursor, now);
+            if (first === null)
+                continue;
+            let endIndex = cursor + first.partsConsumed - 1;
+            let dateStart = null;
+            let dueDate = first.date;
+            let isSoft = isSoftPrefix;
+            // Span continuation: `<date>~<date>`
+            if (endIndex + 1 < parts.length && parts[endIndex + 1] === '~') {
+                const second = tryParseSingleDate(parts, endIndex + 2, now);
+                if (second !== null) {
+                    if (isSoftPrefix) {
+                        // `due ~start~end` is malformed; reject (treat as if span didn't parse).
+                        // Fall through with the first date only.
+                    }
+                    else {
+                        dateStart = first.date;
+                        dueDate = second.date;
+                        endIndex = endIndex + 1 + second.partsConsumed; // skip the `~` and the second date's parts
+                        isSoft = true; // span implies soft
                     }
-                    const finalAbsoluteEnd = groupStart + 1 + endPos;
-                    tokens.push({
-                        type: 'due',
-                        value: textDateResult.date,
-                        raw: rawParts.join(' '),
-                        start: absoluteStart,
-                        end: finalAbsoluteEnd,
-                        soft: isSoft || undefined,
-                    });
-                    continue;
                 }
             }
+            // Mark all consumed including 'due' and optional leading '~'
+            for (let j = i; j <= endIndex; j++)
+                skipIndices.add(j);
+            const raw = content.slice(partOffsets[i], partEndOffset(endIndex));
+            tokens.push({
+                type: 'due',
+                value: dueDate,
+                raw,
+                start: absStart(i),
+                end: absEnd(endIndex),
+                soft: isSoft || undefined,
+                dateStart: dateStart ?? undefined,
+            });
             continue;
         }
-        // Check for priority
+        // priority
         const priorityMatch = part.match(PRIORITY_REGEX);
         if (priorityMatch) {
             tokens.push({
                 type: 'priority',
                 value: parseInt(priorityMatch[1], 10),
                 raw: part,
-                start: absoluteStart,
-                end: absoluteEnd,
+                start: absStart(i),
+                end: absEnd(i),
             });
             continue;
         }
-        // Check for estimate
+        // estimate (must run before relative-date so bare 2d/2h/2m stay as estimates)
         const estimateMatch = part.match(ESTIMATE_REGEX);
         if (estimateMatch) {
             const num = parseInt(estimateMatch[1], 10);
@@ -332,118 +371,98 @@ export function parseTokensInParenGroup(content, groupStart) {
                 type: 'estimate',
                 value: minutes,
                 raw: part,
-                start: absoluteStart,
-                end: absoluteEnd,
+                start: absStart(i),
+                end: absEnd(i),
             });
             continue;
         }
-        // Check for progress states
+        // progress (single word)
         const partLower = part.toLowerCase();
-        // Check for single-word progress states: done, deleted, blocked
         if (PROGRESS_STATES.has(partLower)) {
             tokens.push({
                 type: 'progress',
                 value: partLower,
                 raw: part,
-                start: absoluteStart,
-                end: absoluteEnd,
+                start: absStart(i),
+                end: absEnd(i),
             });
             continue;
         }
-        // Check for multi-word progress state: "in progress"
-        if (partLower === 'in') {
-            const remainingParts = parts.slice(i + 1);
-            if (remainingParts.length > 0 && remainingParts[0].toLowerCase() === 'progress') {
-                const nextPartStart = content.indexOf(remainingParts[0], currentPos);
-                const nextAbsoluteEnd = groupStart + 1 + nextPartStart + remainingParts[0].length;
-                tokens.push({
-                    type: 'progress',
-                    value: 'in progress',
-                    raw: `${part} ${remainingParts[0]}`,
-                    start: absoluteStart,
-                    end: nextAbsoluteEnd,
-                });
-                skipIndices.add(i + 1);
-                continue;
-            }
+        // progress: "in progress" — but `in` also introduces relative dates.
+        // Disambiguate by lookahead: if next part is "progress", it's a progress state.
+        if (partLower === 'in' && i + 1 < parts.length && parts[i + 1].toLowerCase() === 'progress') {
+            tokens.push({
+                type: 'progress',
+                value: 'in progress',
+                raw: `${part} ${parts[i + 1]}`,
+                start: absStart(i),
+                end: absEnd(i + 1),
+            });
+            skipIndices.add(i + 1);
+            continue;
         }
-        // Check for hashtags
+        // hashtag
         const hashtagMatch = part.match(HASHTAG_REGEX);
         if (hashtagMatch) {
             tokens.push({
                 type: 'hashtag',
                 value: hashtagMatch[1].toLowerCase(),
                 raw: part,
-                start: absoluteStart,
-                end: absoluteEnd,
+                start: absStart(i),
+                end: absEnd(i),
             });
             continue;
         }
-        // Check for standalone dates (without "due" prefix)
-        // Check for soft date prefix (~)
-        let isSoftStandalone = false;
-        let partToCheck = part;
-        if (part.startsWith('~')) {
-            isSoftStandalone = true;
-            partToCheck = part.slice(1);
-        }
-        // First try standard date formats (single part)
-        const standaloneDateResult = parseDate(partToCheck);
-        if (standaloneDateResult.valid) {
-            tokens.push({
-                type: 'due',
-                value: standaloneDateResult.date,
-                raw: part,
-                start: absoluteStart,
-                end: absoluteEnd,
-                soft: isSoftStandalone || undefined,
-            });
-            continue;
-        }
-        // Try text date formats starting with this part (Month Day [Year] or Day Month [Year])
-        const remainingPartsForDate = parts.slice(i);
-        // For text dates, check if first part starts with ~
-        let datePartsForTextParsing = remainingPartsForDate;
-        if (isSoftStandalone) {
-            datePartsForTextParsing = [partToCheck, ...remainingPartsForDate.slice(1)];
-        }
-        const standaloneTextDateResult = parseTextDate(datePartsForTextParsing);
-        if (standaloneTextDateResult.valid) {
-            // Calculate the end position
-            let rawParts = [];
-            let endPos = partStart;
-            for (let j = 0; j < standaloneTextDateResult.partsConsumed; j++) {
-                rawParts.push(remainingPartsForDate[j]);
-                if (j > 0)
-                    skipIndices.add(i + j);
-                endPos = content.indexOf(remainingPartsForDate[j], endPos) + remainingPartsForDate[j].length;
+        // Standalone date / relative / span
+        const first = tryParseSingleDate(parts, i, now);
+        if (first !== null) {
+            let endIndex = i + first.partsConsumed - 1;
+            let dateStart = null;
+            let dueDate = first.date;
+            let isSoft = false;
+            if (endIndex + 1 < parts.length && parts[endIndex + 1] === '~') {
+                const second = tryParseSingleDate(parts, endIndex + 2, now);
+                if (second !== null) {
+                    dateStart = first.date;
+                    dueDate = second.date;
+                    endIndex = endIndex + 1 + second.partsConsumed;
+                    isSoft = true;
+                }
             }
-            const finalAbsoluteEnd = groupStart + 1 + endPos;
+            for (let j = i; j <= endIndex; j++)
+                skipIndices.add(j);
+            const raw = content.slice(partOffsets[i], partEndOffset(endIndex));
             tokens.push({
                 type: 'due',
-                value: standaloneTextDateResult.date,
-                raw: rawParts.join(' '),
-                start: absoluteStart,
-                end: finalAbsoluteEnd,
-                soft: isSoftStandalone || undefined,
+                value: dueDate,
+                raw,
+                start: absStart(i),
+                end: absEnd(endIndex),
+                soft: isSoft || undefined,
+                dateStart: dateStart ?? undefined,
             });
             continue;
         }
     }
     return {
         start: groupStart,
-        end: groupStart + content.length + 2, // +2 for parens
+        end: groupStart + content.length + 2,
         content,
         tokens,
         hasRecognizedTokens: tokens.length > 0,
     };
 }
-export function extractParenGroups(line, lineStart) {
+export function extractParenGroups(line, _lineStart, now) {
+    if (!now)
+        now = resolveNow(undefined, undefined);
+    return extractParenGroupsImpl(line, _lineStart, now);
+}
+function extractParenGroupsImpl(line, _lineStart, now) {
     const groups = [];
     let i = 0;
     while (i < line.length) {
         if (line[i] === '(') {
-            const start = i; // Position within the line (content)
+            const start = i;
             let depth = 1;
             i++;
             while (i < line.length && depth > 0) {
@@ -455,10 +474,9 @@ export function extractParenGroups(line, lineStart) {
             }
             if (depth === 0) {
                 const content = line.slice(start + 1, i - 1);
-                // Pass the relative position within content string
-                const group = parseTokensInParenGroup(content, start);
-                group.start = start; // Store relative position
-                group.end = i; // Store relative position
+                const group = parseTokensInParenGroup(content, start, now);
+                group.start = start;
+                group.end = i;
                 groups.push(group);
             }
         }
@@ -469,23 +487,19 @@ export function extractParenGroups(line, lineStart) {
     return groups;
 }
 function buildTitleText(rawText, groups) {
-    // Remove groups that have recognized tokens
-    // Process in reverse order to maintain correct positions
     const sortedGroups = [...groups]
         .filter(g => g.hasRecognizedTokens)
         .sort((a, b) => b.start - a.start);
     let result = rawText;
     for (const group of sortedGroups) {
-        const before = result.slice(0, group.start);
-        const after = result.slice(group.end);
-        result = before + after;
+        result = result.slice(0, group.start) + result.slice(group.end);
     }
-    // Clean up extra whitespace
     return result.replace(/\s+/g, ' ').trim();
 }
 function buildMetadata(groups) {
     const metadata = {
         due: null,
+        due_start: null,
         due_soft: null,
         priority: null,
         estimate_minutes: null,
@@ -493,15 +507,13 @@ function buildMetadata(groups) {
         hashtags: [],
         effective_hashtags: [],
     };
-    // Collect all tokens from all groups
     const allTokens = groups.flatMap(g => g.tokens);
-    // Collect unique hashtags (sorted alphabetically)
     const hashtagSet = new Set();
-    // Last occurrence wins for non-hashtag tokens
     for (const token of allTokens) {
         switch (token.type) {
             case 'due':
                 metadata.due = token.value;
+                metadata.due_start = token.dateStart ?? null;
                 metadata.due_soft = token.soft ?? null;
                 break;
             case 'priority':
@@ -518,35 +530,30 @@ function buildMetadata(groups) {
                 break;
         }
     }
-    // Store sorted unique hashtags
     metadata.hashtags = [...hashtagSet].sort();
     return metadata;
 }
-export function parse(text) {
+export function parse(text, options) {
+    const now = resolveNow(options?.now, options?.settings);
     const lines = text.split('\n');
     const items = [];
     const warnings = [];
     let nextId = 0;
     let offset = 0;
-    // Stack to track current context: [itemId, indentLevel]
     const listStack = [];
     let currentHeadingId = null;
     const rootIds = [];
-    // Map id -> children for building the tree
     const childrenMap = new Map();
-    // First pass: identify all items and their basic info
     for (let lineNum = 0; lineNum < lines.length; lineNum++) {
         const line = lines[lineNum];
         const lineStart = offset;
         const lineEnd = offset + line.length;
-        // Check for heading
         const headingMatch = line.match(HEADING_REGEX);
         if (headingMatch) {
             const level = headingMatch[1].length;
             const content = headingMatch[2];
-            // Close any open list context
             listStack.length = 0;
-            const groups = extractParenGroups(content, lineStart + headingMatch[1].length + 1);
+            const groups = extractParenGroups(content, lineStart + headingMatch[1].length + 1, now);
             const titleText = buildTitleText(content, groups);
             const metadata = buildMetadata(groups);
             const id = String(nextId++);
@@ -571,17 +578,15 @@ export function parse(text) {
             offset = lineEnd + 1;
             continue;
         }
-        // Check for list item
         const listMatch = line.match(LIST_ITEM_REGEX);
         if (listMatch) {
             const indent = listMatch[1].length;
             const marker = listMatch[2];
             const content = listMatch[3];
             const contentStart = lineStart + indent + marker.length + 1;
-            const groups = extractParenGroups(content, contentStart);
+            const groups = extractParenGroups(content, contentStart, now);
             const titleText = buildTitleText(content, groups);
             const metadata = buildMetadata(groups);
-            // Determine marker type and sequence number
             const isNumbered = /^\d+\.$/.test(marker);
             const markerType = isNumbered ? 'numbered' : 'bullet';
             const sequenceNumber = isNumbered ? parseInt(marker.slice(0, -1), 10) : undefined;
@@ -603,29 +608,23 @@ export function parse(text) {
             };
             items.push(item);
             childrenMap.set(id, []);
-            // Determine parent
-            // Pop items from stack that are at same or greater indent
             while (listStack.length > 0 && listStack[listStack.length - 1].indent >= indent) {
                 listStack.pop();
             }
             if (listStack.length > 0) {
-                // Parent is the top of the stack
                 const parentId = listStack[listStack.length - 1].id;
                 childrenMap.get(parentId).push(id);
             }
             else if (currentHeadingId !== null) {
-                // Parent is current heading
                 childrenMap.get(currentHeadingId).push(id);
             }
             else {
-                // No parent, it's a root
                 rootIds.push(id);
             }
             listStack.push({ id, indent });
             offset = lineEnd + 1;
             continue;
         }
-        // Not a heading or list item - could be a comment or blank line
         offset = lineEnd + 1;
     }
     // Second pass: collect comments
@@ -637,7 +636,6 @@ export function parse(text) {
         const line = lines[lineNum];
         const lineStart = offset;
         const lineEnd = offset + line.length;
-        // Check if this line starts a new item
         const headingMatch = line.match(HEADING_REGEX);
         const listMatch = line.match(LIST_ITEM_REGEX);
         if (headingMatch || listMatch) {
@@ -647,17 +645,12 @@ export function parse(text) {
             offset = lineEnd + 1;
             continue;
         }
-        // Check for blank line
         if (line.trim() === '') {
-            if (hasStartedComments) {
-                // Blank line after comments started = stop collecting
+            if (hasStartedComments)
                 blankAfterCommentStart = true;
-            }
-            // Blank line before comments started (e.g., after heading) = ignore
             offset = lineEnd + 1;
             continue;
         }
-        // Non-blank, non-item line - potential comment
         if (currentItemIndex >= 0 && !blankAfterCommentStart) {
             const currentItem = items[currentItemIndex];
             currentItem.comments.push(line.trim());
@@ -666,11 +659,9 @@ export function parse(text) {
         }
         offset = lineEnd + 1;
     }
-    // Build children arrays and compute subtree spans
     for (const item of items) {
         item.children = childrenMap.get(item.id) || [];
     }
-    // Compute subtree spans (post-order traversal)
     function computeSubtreeSpan(id) {
         const item = items.find(i => i.id === id);
         let end = item.item_span[1];
@@ -684,17 +675,11 @@ export function parse(text) {
     for (const rootId of rootIds) {
         computeSubtreeSpan(rootId);
     }
-    // Update root_ids to only include top-level items
     const actualRootIds = items
-        .filter(item => {
-        // Check if this item is a child of any other item
-        return !items.some(other => other.children.includes(item.id));
-    })
+        .filter(item => !items.some(other => other.children.includes(item.id)))
         .map(item => item.id);
-    // Compute effective_hashtags through inheritance (pre-order traversal)
     function computeEffectiveHashtags(id, parentEffectiveHashtags) {
         const item = items.find(i => i.id === id);
-        // Merge parent's effective_hashtags with own hashtags, deduplicate and sort
         const combined = new Set([...parentEffectiveHashtags, ...item.metadata.hashtags]);
         item.metadata.effective_hashtags = [...combined].sort();
         for (const childId of item.children) {
@@ -705,10 +690,7 @@ export function parse(text) {
         computeEffectiveHashtags(rootId, []);
     }
     return {
-        ast: {
-            items,
-            root_ids: actualRootIds,
-        },
+        ast: { items, root_ids: actualRootIds },
         warnings,
     };
 }