chatstrata 0.1.0__py3-none-any.whl

chatstrata/__init__.py

@@ -0,0 +1,7 @@
+"""chatstrata - a personal, queryable archive of your AI conversations."""
+
+from chatstrata.core.migrations import LATEST_VERSION as SCHEMA_VERSION
+
+__version__ = "0.1.0"
+
+__all__ = ["SCHEMA_VERSION", "__version__"]

chatstrata/analysis/__init__.py

@@ -0,0 +1,24 @@
+"""Analysis query helpers."""
+
+from __future__ import annotations
+
+from pathlib import Path
+
+_QUERIES_DIR = Path(__file__).parent / "queries"
+
+
+def load_query(name: str) -> str:
+    """Load a .sql file from the queries directory by name (without extension)."""
+    path = _QUERIES_DIR / f"{name}.sql"
+    if not path.exists():
+        raise FileNotFoundError(f"Query file not found: {path}")
+    return path.read_text()
+
+
+def build_source_filter(
+    source: str | None, *, column: str = "c.source_id"
+) -> tuple[str, list]:
+    """Return a (sql_fragment, params) tuple for optional source filtering."""
+    if source is None:
+        return "", []
+    return f"AND {column} = ?", [source]

chatstrata/analysis/cli.py

@@ -0,0 +1,145 @@
+"""Analyze subcommands for chatstrata."""
+
+from __future__ import annotations
+
+import json
+
+import click
+
+from chatstrata.analysis import build_source_filter, load_query
+from chatstrata.core.db import connect, resolve_db_path
+
+
+def _output(cols: list[str], rows: list[tuple], as_json: bool) -> None:
+    """Output results as either a formatted table or JSON."""
+    if as_json:
+        out = [dict(zip(cols, row, strict=False)) for row in rows]
+        click.echo(json.dumps(out, default=str, indent=2))
+        return
+    if not rows:
+        click.echo("No data.")
+        return
+    widths = [
+        max(len(c), *(len(str(v) if v is not None else "") for v in (r[i] for r in rows)))
+        for i, c in enumerate(cols)
+    ]
+    click.echo("  ".join(c.ljust(w) for c, w in zip(cols, widths)))
+    click.echo("  ".join("-" * w for w in widths))
+    for row in rows:
+        click.echo("  ".join(str(v if v is not None else "").ljust(w) for v, w in zip(row, widths)))
+
+
+@click.group()
+def analyze() -> None:
+    """Analyze your conversation archive."""
+
+
+@analyze.command()
+@click.option(
+    "--by", "granularity", type=click.Choice(["day", "week", "month"]),
+    default="month", help="Time granularity (default: month).",
+)
+@click.option("--source", default=None, help="Filter to a specific source.")
+@click.option("--db", default=None, help="Override the database path.")
+@click.option("--json", "as_json", is_flag=True, help="Output as JSON.")
+def activity(granularity: str, source: str | None, db: str | None, as_json: bool) -> None:
+    """Messages over time, grouped by period."""
+    sql_template = load_query("activity")
+    source_filter, params = build_source_filter(source)
+    sql = sql_template.format(granularity=granularity, source_filter=source_filter)
+
+    conn = connect(resolve_db_path(db))
+    try:
+        result = conn.execute(sql, params)
+        cols = [d[0] for d in result.description]
+        rows = result.fetchall()
+        _output(cols, rows, as_json)
+    finally:
+        conn.close()
+
+
+@analyze.command()
+@click.option("--source", default=None, help="Filter to a specific source (e.g. claude_code).")
+@click.option("--db", default=None, help="Override the database path.")
+@click.option("--json", "as_json", is_flag=True, help="Output as JSON.")
+def tools(source: str | None, db: str | None, as_json: bool) -> None:
+    """Tool usage frequency."""
+    sql_template = load_query("tools")
+    source_filter, params = build_source_filter(source, column="source_id")
+    sql = sql_template.format(source_filter=source_filter)
+
+    conn = connect(resolve_db_path(db))
+    try:
+        result = conn.execute(sql, params)
+        cols = [d[0] for d in result.description]
+        rows = result.fetchall()
+        _output(cols, rows, as_json)
+    finally:
+        conn.close()
+
+
+@analyze.command()
+@click.option("--longest", "longest_n", type=int, default=None, help="Show N longest conversations.")
+@click.option("--shortest", "shortest_n", type=int, default=None, help="Show N shortest conversations.")
+@click.option("--db", default=None, help="Override the database path.")
+@click.option("--json", "as_json", is_flag=True, help="Output as JSON.")
+def conversations(
+    longest_n: int | None, shortest_n: int | None, db: str | None, as_json: bool,
+) -> None:
+    """Conversation length statistics."""
+    if longest_n and shortest_n:
+        raise click.UsageError("Specify --longest or --shortest, not both.")
+
+    order = "DESC"
+    limit = 20
+    if shortest_n:
+        order = "ASC"
+        limit = shortest_n
+    elif longest_n:
+        limit = longest_n
+
+    sql_template = load_query("conversations")
+    sql = sql_template.format(order=order, limit=limit)
+
+    conn = connect(resolve_db_path(db))
+    try:
+        result = conn.execute(sql)
+        cols = [d[0] for d in result.description]
+        rows = result.fetchall()
+        _output(cols, rows, as_json)
+    finally:
+        conn.close()
+
+
+@analyze.command()
+@click.option("--db", default=None, help="Override the database path.")
+@click.option("--json", "as_json", is_flag=True, help="Output as JSON.")
+def models(db: str | None, as_json: bool) -> None:
+    """Model usage breakdown."""
+    sql = load_query("models")
+
+    conn = connect(resolve_db_path(db))
+    try:
+        result = conn.execute(sql)
+        cols = [d[0] for d in result.description]
+        rows = result.fetchall()
+        _output(cols, rows, as_json)
+    finally:
+        conn.close()
+
+
+@analyze.command()
+@click.option("--db", default=None, help="Override the database path.")
+@click.option("--json", "as_json", is_flag=True, help="Output as JSON.")
+def projects(db: str | None, as_json: bool) -> None:
+    """Per-project conversation counts (Claude Code)."""
+    sql = load_query("projects")
+
+    conn = connect(resolve_db_path(db))
+    try:
+        result = conn.execute(sql)
+        cols = [d[0] for d in result.description]
+        rows = result.fetchall()
+        _output(cols, rows, as_json)
+    finally:
+        conn.close()

chatstrata/analysis/queries/activity.sql

@@ -0,0 +1,12 @@
+SELECT
+    s.id AS source,
+    date_trunc('{granularity}', m.created_at) AS period,
+    COUNT(*) AS messages,
+    COUNT(DISTINCT m.conversation_id) AS conversations
+FROM messages m
+JOIN conversations c ON c.id = m.conversation_id
+JOIN sources s ON s.id = c.source_id
+WHERE m.created_at IS NOT NULL
+  {source_filter}
+GROUP BY source, period
+ORDER BY period DESC, source

chatstrata/analysis/queries/conversations.sql

@@ -0,0 +1,11 @@
+SELECT
+    c.title,
+    s.id AS source,
+    c.project,
+    c.message_count,
+    c.started_at,
+    c.ended_at
+FROM conversations c
+JOIN sources s ON s.id = c.source_id
+ORDER BY c.message_count {order}
+LIMIT {limit}

chatstrata/analysis/queries/example_queries.sql

@@ -0,0 +1,129 @@
+-- Example chatstrata queries. Use as: chatstrata query "$(cat path/to/query.sql)"
+-- or paste into a DuckDB shell connected to your chatstrata.duckdb file.
+
+-- ============================================================
+-- 1. Activity over time
+-- ============================================================
+
+-- Messages per month, by source
+SELECT
+    s.id AS source,
+    date_trunc('month', m.created_at) AS month,
+    COUNT(*) AS messages
+FROM messages m
+JOIN conversations c ON c.id = m.conversation_id
+JOIN sources s ON s.id = c.source_id
+WHERE m.created_at IS NOT NULL
+GROUP BY source, month
+ORDER BY month DESC, source;
+
+-- ============================================================
+-- 2. Conversation length distribution
+-- ============================================================
+
+SELECT
+    CASE
+        WHEN message_count < 5 THEN 'very short (< 5)'
+        WHEN message_count < 20 THEN 'short (5-19)'
+        WHEN message_count < 50 THEN 'medium (20-49)'
+        ELSE 'long (50+)'
+    END AS length_bucket,
+    COUNT(*) AS conversations
+FROM conversations
+GROUP BY length_bucket
+ORDER BY conversations DESC;
+
+-- ============================================================
+-- 3. Every bash command via Claude Code, grouped by project
+-- ============================================================
+
+SELECT
+    c.project,
+    cb.payload->'input'->>'command' AS command,
+    COUNT(*) AS times_run
+FROM content_blocks cb
+JOIN messages m ON m.id = cb.message_id
+JOIN conversations c ON c.id = m.conversation_id
+WHERE cb.type = 'tool_use'
+  AND cb.tool_name = 'Bash'
+  AND c.source_id = 'claude_code'
+GROUP BY c.project, command
+ORDER BY times_run DESC
+LIMIT 50;
+
+-- ============================================================
+-- 4. Tool usage frequency
+-- ============================================================
+
+SELECT
+    tool_name,
+    COUNT(*) AS calls,
+    COUNT(DISTINCT conversation_id) AS conversations
+FROM tool_calls
+GROUP BY tool_name
+ORDER BY calls DESC;
+
+-- ============================================================
+-- 5. Word count of user prompts over time
+-- ============================================================
+
+-- Rough proxy for "how have I been prompting"
+SELECT
+    date_trunc('week', m.created_at) AS week,
+    AVG(length(cb.text) - length(replace(cb.text, ' ', ''))) AS avg_word_count
+FROM content_blocks cb
+JOIN messages m ON m.id = cb.message_id
+WHERE m.role = 'user'
+  AND cb.type = 'text'
+  AND cb.text IS NOT NULL
+GROUP BY week
+ORDER BY week;
+
+-- ============================================================
+-- 6. Recent conversations matching a keyword
+-- ============================================================
+
+SELECT
+    c.started_at,
+    s.id AS source,
+    c.project,
+    c.title
+FROM conversations c
+JOIN sources s ON s.id = c.source_id
+WHERE EXISTS (
+    SELECT 1 FROM messages m
+    JOIN content_blocks cb ON cb.message_id = m.id
+    WHERE m.conversation_id = c.id
+      AND cb.text ILIKE '%grandma%'  -- replace with your keyword
+)
+ORDER BY c.started_at DESC
+LIMIT 20;
+
+-- ============================================================
+-- 7. Most-used models
+-- ============================================================
+
+SELECT model, COUNT(*) AS messages
+FROM messages
+WHERE model IS NOT NULL
+GROUP BY model
+ORDER BY messages DESC;
+
+-- ============================================================
+-- 8. Thinking blocks: how often does the model reason explicitly?
+-- ============================================================
+
+SELECT
+    date_trunc('month', m.created_at) AS month,
+    SUM(CASE WHEN cb.type = 'thinking' THEN 1 ELSE 0 END) AS thinking_blocks,
+    COUNT(DISTINCT m.id) AS assistant_messages,
+    ROUND(
+        100.0 * SUM(CASE WHEN cb.type = 'thinking' THEN 1 ELSE 0 END)
+              / NULLIF(COUNT(DISTINCT m.id), 0),
+        1
+    ) AS pct
+FROM messages m
+LEFT JOIN content_blocks cb ON cb.message_id = m.id
+WHERE m.role = 'assistant'
+GROUP BY month
+ORDER BY month DESC;

chatstrata/analysis/queries/models.sql

@@ -0,0 +1,8 @@
+SELECT
+    model,
+    COUNT(*) AS messages,
+    COUNT(DISTINCT m.conversation_id) AS conversations
+FROM messages m
+WHERE model IS NOT NULL
+GROUP BY model
+ORDER BY messages DESC

chatstrata/analysis/queries/projects.sql

@@ -0,0 +1,11 @@
+SELECT
+    c.project,
+    COUNT(*) AS conversations,
+    SUM(c.message_count) AS total_messages,
+    MIN(c.started_at) AS earliest,
+    MAX(c.ended_at) AS latest
+FROM conversations c
+WHERE c.source_id = 'claude_code'
+  AND c.project IS NOT NULL
+GROUP BY c.project
+ORDER BY conversations DESC

chatstrata/analysis/queries/tools.sql

@@ -0,0 +1,9 @@
+SELECT
+    tool_name,
+    COUNT(*) AS calls,
+    COUNT(DISTINCT conversation_id) AS conversations
+FROM tool_calls
+WHERE tool_name IS NOT NULL
+  {source_filter}
+GROUP BY tool_name
+ORDER BY calls DESC