transcript-viewer 0.7.2__tar.gz → 0.8.1__tar.gz

{transcript_viewer-0.7.2 → transcript_viewer-0.8.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: transcript-viewer
-Version: 0.7.2
+Version: 0.8.1
 Summary: Browse agent transcripts in a local, dependency-free web viewer.
 License: MIT
 Requires-Python: >=3.12

{transcript_viewer-0.7.2 → transcript_viewer-0.8.1}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "transcript-viewer"
-version = "0.7.2"
+version = "0.8.1"
 description = "Browse agent transcripts in a local, dependency-free web viewer."
 readme = "README.md"
 requires-python = ">=3.12"

{transcript_viewer-0.7.2 → transcript_viewer-0.8.1}/src/transcript_viewer/cli.py

@@ -49,7 +49,12 @@ def cmd_view(args: argparse.Namespace) -> int:
         # An explicit path that holds nothing is a mistake worth reporting; an
         # empty library is not, so this only guards the argument.
         if not entries:
-            print(f"transcript-viewer: nothing convertible in {path}", file=sys.stderr)
+            print(
+                f"transcript-viewer: no transcript in {path}. "
+                f"{corpus.describe_contents(path)}, and none of it is a log "
+                f"this can read.",
+                file=sys.stderr,
+            )
             return 1
     else:
         # Only what has been added deliberately. Scanning a machine because the

{transcript_viewer-0.7.2 → transcript_viewer-0.8.1}/src/transcript_viewer/corpus.py

@@ -9,6 +9,7 @@ something actually asks for that trajectory.
 from __future__ import annotations
 
 import hashlib
+from collections import Counter
 import re
 import json
 from dataclasses import asdict, dataclass
@@ -367,3 +368,32 @@ def load(path: Path | None = None) -> list[Entry]:
         except TypeError:
             continue
     return entries
+
+
+def describe_contents(root: Path, limit: int = 4000) -> str:
+    """What is inside a path, said in a sentence.
+
+    Reached only when nothing in it converted. "Nothing convertible" on its own
+    invites the reader to doubt the tool rather than the file — one real archive
+    held 1,952 files and an audit that referenced a transcript which had not
+    been shipped with it, and the answer gave no way to tell that from a bug.
+    """
+    if root.is_file():
+        return f"{root.name} ({root.stat().st_size:,} bytes)"
+
+    kinds: Counter[str] = Counter()
+    total = 0
+    for item in root.rglob("*"):
+        if not item.is_file():
+            continue
+        total += 1
+        if total > limit:
+            break
+        kinds[item.suffix.lower() or "no extension"] += 1
+
+    if not total:
+        return "it is empty"
+
+    common = ", ".join(f"{n:,} {ext}" for ext, n in kinds.most_common(4))
+    more = "+" if total > limit else ""
+    return f"{total:,}{more} files inside — {common}"

{transcript_viewer-0.7.2 → transcript_viewer-0.8.1}/src/transcript_viewer/page.html

@@ -431,6 +431,26 @@ pre{background:var(--sunk);border-radius:7px;padding:10px;overflow:auto;font-siz
 .sh-op{color:var(--sh-op);font-weight:500}
 .sh-var{color:var(--sh-var)}
 .sh-com{color:var(--sh-com);font-style:italic}
+
+/* A file, shown as a file. The numbers sit in a gutter and are not selectable,
+   so copying the block gives back the file rather than the listing of it. */
+.fpath{font:11.5px/1.5 var(--font-mono);color:var(--muted);margin-bottom:6px;
+  overflow-wrap:anywhere}
+.fmeta{font:11px/1.5 var(--font-mono);color:var(--faint);margin-top:6px}
+.filelines,.filebody{background:var(--bg);border-radius:7px;padding:10px 12px;
+  margin:0;max-height:340px;overflow:auto;font:12px/1.6 var(--font-mono);
+  color:var(--ink);white-space:pre-wrap;overflow-wrap:anywhere}
+.ln{display:inline-block;width:3.2em;margin-right:.8em;text-align:right;
+  color:var(--faint);user-select:none;-webkit-user-select:none}
+/* What a change replaced, and what it put there. */
+.epair{display:grid;gap:8px}
+.eside{border-radius:7px;padding:8px 10px;background:var(--bg);
+  font:12px/1.6 var(--font-mono);white-space:pre-wrap;overflow-wrap:anywhere;
+  max-height:260px;overflow:auto}
+.eside b{display:block;font:11px/1.5 var(--font-mono);font-weight:500;
+  margin-bottom:4px;letter-spacing:.04em;text-transform:uppercase}
+.eside.out b{color:var(--danger)}
+.eside.in b{color:var(--ok)}
 /* What the call was for, beside what it was called. */
 .tdesc{color:var(--muted);font-size:12px;overflow:hidden;text-overflow:ellipsis;
   white-space:nowrap;flex:1;min-width:0}
@@ -889,6 +909,44 @@ function sh(src) {
 
 const shBlock = (command) => `<pre class="sh">${autolink(sh(command))}</pre>`;
 
+/* Where a file is, as something you can open.
+
+   Linked whole rather than through autolink: this value is a path in its
+   entirety, where autolink has to guess boundaries in prose and stops at the
+   first space. A real one here — "/Users/…/Redwood Research/…" — linked only
+   as far as "Redwood", which pointed the link at a directory that is not the
+   file. PATH_EXACT is the same test the JSON view uses for a quoted value. */
+const fileRef = (path) => {
+  const text = String(path ?? "");
+  if (!text) return "";
+  const shown = PATH_EXACT.test(text) ? pathAnchor(text, esc(text)) : esc(text);
+  return `<div class="fpath">${shown}</div>`;
+};
+
+const countOf = (text) => {
+  const s = String(text ?? "");
+  const lines = s ? s.split("\n").length : 0;
+  return `${num(lines)} line${lines === 1 ? "" : "s"} · ${bytes(s.length)}`;
+};
+
+/* A file read back arrives numbered, one "12\tline" per line. Put the numbers
+   in a gutter so the file reads as a file, and leave them unselectable so that
+   copying gives the contents rather than the listing. */
+function numbered(text) {
+  const rows = String(text ?? "")
+    .split("\n")
+    .map((line) => {
+      const m = /^(\s*\d+)\t([\s\S]*)$/.exec(line);
+      return m
+        ? `<span class="ln">${esc(m[1].trim())}</span>${esc(m[2])}`
+        : `<span class="ln"></span>${esc(line)}`;
+    });
+  return `<pre class="filelines">${autolink(rows.join("\n"))}</pre>`;
+}
+
+const fileBlock = (text) =>
+  `<pre class="filebody">${autolink(esc(String(text ?? "")))}</pre>`;
+
 /* What a call is worth showing as, when it is not worth showing as JSON. Only
    the shapes that carry a command or an address get a body of their own; the
    rest gain a title, which is the part that was missing everywhere. */
@@ -910,9 +968,37 @@ const TOOLVIEW = {
   },
   WebSearch: { title: (a) => a.query || "" },
   // The end of a path is the part worth reading; the full one is on hover.
-  Read: { title: (a) => short(a.file_path) },
-  Write: { title: (a) => short(a.file_path) },
-  Edit: { title: (a) => short(a.file_path) },
+  Read: {
+    title: (a) => short(a.file_path),
+    body: (a) =>
+      fileRef(a.file_path) +
+      (a.offset || a.limit || a.pages
+        ? `<div class="fmeta">${
+            a.pages
+              ? `pages ${esc(String(a.pages))}`
+              : `from line ${num(a.offset || 1)}${a.limit ? `, ${num(a.limit)} lines` : ""}`
+          }</div>`
+        : ""),
+    // What came back is the file, numbered as the reader saw it.
+    output: (content) => numbered(content),
+  },
+  Write: {
+    title: (a) => short(a.file_path),
+    body: (a) =>
+      fileRef(a.file_path) +
+      fileBlock(a.content) +
+      `<div class="fmeta">${countOf(a.content)}</div>`,
+  },
+  Edit: {
+    title: (a) => short(a.file_path),
+    body: (a) =>
+      fileRef(a.file_path) +
+      `<div class="epair">
+         <div class="eside out"><b>replaced</b>${autolink(esc(a.old_string ?? ""))}</div>
+         <div class="eside in"><b>with</b>${autolink(esc(a.new_string ?? ""))}</div>
+       </div>` +
+      (a.replace_all ? `<div class="fmeta">every occurrence</div>` : ""),
+  },
   NotebookEdit: { title: (a) => short(a.notebook_path) },
   Skill: { title: (a) => a.skill || "" },
   Task: { title: (a) => a.description || "" },
@@ -943,7 +1029,11 @@ function toolBody(id) {
     `<div class="io in"><span class="iol">call</span>${shown}</div>` +
     (r && r.content
       ? `<div class="io out"><span class="iol">output</span>${
-          typeof r.content === "string" ? pre(r.content) : body(r.content)
+          typeof r.content !== "string"
+            ? body(r.content)
+            : !toolIsRaw(id) && view && view.output
+              ? view.output(r.content, c.arguments || {})
+              : pre(r.content)
         }</div>`
       : "") +
     (aiOn() ? callSummary(c.tool_call_id) : "")

{transcript_viewer-0.7.2 → transcript_viewer-0.8.1}/src/transcript_viewer/viewer.py

@@ -27,6 +27,7 @@ from urllib.parse import parse_qs, unquote, urlparse
 from atif_make.atif import ContentPart, Trajectory
 from . import corpus
 from atif_make.container import is_container, open_container
+from atif_make.detect import FORMATS
 from atif_make.convert import convert
 from transcript_viewer.corpus import Entry, scan
 
@@ -615,9 +616,19 @@ class _Handler(BaseHTTPRequestHandler):
             self._json({"error": str(exc)}, 400)
             return
         if not found:
-            # Nothing usable in it — do not keep the copy around.
+            # Nothing usable in it — do not keep the copy around. Say what was
+            # in it on the way out: an answer of "nothing convertible" with
+            # nothing behind it reads as the tool failing rather than the file
+            # not holding what was hoped.
+            inside = corpus.describe_contents(target)
             shutil.rmtree(home, ignore_errors=True)
-            self._json({"error": f"nothing convertible in {name}"}, 415)
+            self._json(
+                {
+                    "error": f"No transcript in {name}. {inside}, and none of it "
+                    f"is a log this can read ({', '.join(FORMATS)})."
+                },
+                415,
+            )
             return
 
         with self.lock:

{transcript_viewer-0.7.2 → transcript_viewer-0.8.1}/tests/page.test.js

@@ -1673,3 +1673,103 @@ test("a real 20,000-character command draws, and quickly", () => {
   assert.ok(html.includes("class=\"sh\""), "it should render");
   assert.ok(took < 2000, `took ${took}ms, which is a page nobody waits for`);
 });
+
+// ---- a file read, written or changed -----------------------------------------
+
+test("a file read back keeps its line numbers in a gutter", () => {
+  const html = run(`return TOOLVIEW.Read.output('1\\tfirst\\n2\\tsecond')`);
+  assert.match(html, /class="filelines"/);
+  assert.match(html, /<span class="ln">1<\/span>first/);
+  assert.match(html, /<span class="ln">2<\/span>second/);
+  assert.ok(!html.includes("\t"), "the tab becomes the gutter, not whitespace");
+});
+
+test("a line that is not numbered still shows", () => {
+  const html = run(`return TOOLVIEW.Read.output('no number here')`);
+  assert.match(html, /no number here/);
+});
+
+test("a read of part of a file says which part", () => {
+  const html = run(`return TOOLVIEW.Read.body({file_path:"/a/b.py",offset:40,limit:120})`);
+  assert.match(html, /from line 40/);
+  assert.match(html, /120 lines/);
+});
+
+test("a whole-file read says nothing about ranges", () => {
+  const html = run(`return TOOLVIEW.Read.body({file_path:"/a/b.py"})`);
+  assert.ok(!/from line/.test(html));
+});
+
+test("a path in a file view still opens in Finder", () => {
+  const html = run(`return TOOLVIEW.Read.body({file_path:"/Users/someone/a.py"})`);
+  assert.match(html, /api\/reveal\?path=/);
+});
+
+test("a write shows what it wrote, and how much", () => {
+  const html = run(
+    `return TOOLVIEW.Write.body({file_path:"/a/b.js",content:"one\\ntwo\\nthree"})`,
+  );
+  assert.match(html, /class="filebody"/);
+  assert.match(html, /three/);
+  assert.match(html, /3 lines/);
+});
+
+test("an edit shows what it replaced and what it put there", () => {
+  const html = run(
+    `return TOOLVIEW.Edit.body({file_path:"/a/b.js",old_string:"before",new_string:"after",replace_all:true})`,
+  );
+  assert.match(html, /class="eside out"[^>]*>.*replaced.*before/s);
+  assert.match(html, /class="eside in"[^>]*>.*with.*after/s);
+  assert.match(html, /every occurrence/);
+});
+
+test("an edit that replaces once does not claim otherwise", () => {
+  const html = run(
+    `return TOOLVIEW.Edit.body({file_path:"/a/b.js",old_string:"x",new_string:"y",replace_all:false})`,
+  );
+  assert.ok(!/every occurrence/.test(html));
+});
+
+test("file contents cannot reach the page as markup", () => {
+  for (const expr of [
+    `TOOLVIEW.Write.body({file_path:"/a",content:"<img src=x onerror=alert(1)>"})`,
+    `TOOLVIEW.Edit.body({file_path:"/a",old_string:"<script>",new_string:"<b>"})`,
+    `TOOLVIEW.Read.output("1\\t<img src=x onerror=alert(1)>")`,
+  ]) {
+    const html = run(`return ${expr}`);
+    assert.ok(!/<img|<script>/.test(html), `escaped badly: ${expr}`);
+  }
+});
+
+test("asking for the recording gives the recording, not the file view", () => {
+  const html = run(`
+    raw = false;
+    CALLS["r1"] = {
+      c:{function_name:"Read",tool_call_id:"r1",arguments:{file_path:"/a/b.py"}},
+      r:{source_call_id:"r1",content:"1\\thello"},
+    };
+    TOOLRAW["r1"] = true;
+    return toolBody("r1");
+  `);
+  assert.match(html, /class="json"/, "the call shows as recorded");
+  assert.ok(!/class="filelines"/.test(html), "and so does the output");
+});
+
+test("a path with a space in it links to the whole path", () => {
+  // A real one: "/Users/girish/Documents/Redwood Research/…" linked only as far
+  // as "Redwood", so the link opened a directory that was not the file.
+  const html = run(
+    `return TOOLVIEW.Read.body({file_path:"/Users/girish/Documents/Redwood Research/app/mentions.py"})`,
+  );
+  assert.match(html, /Redwood%20Research/, "the space must be inside the link");
+  assert.ok(
+    !/<\/a>\s*Research/.test(html),
+    "the link must not stop at the space",
+  );
+});
+
+test("a relative path is shown but not offered as a link", () => {
+  const html = run(`return TOOLVIEW.Read.body({file_path:"src/thing.py"})`);
+  assert.match(html, /src\/thing\.py/);
+  assert.ok(!/api\/reveal/.test(html), "revealing a relative path would guess");
+});

{transcript_viewer-0.7.2 → transcript_viewer-0.8.1}/tests/test_cli.py

@@ -65,7 +65,11 @@ def test_a_path_with_nothing_convertible_reports_rather_than_serving(
 ):
     (tmp_path / "notes.txt").write_text("nothing here")
     assert cli.main([str(tmp_path)]) == 1
-    assert "nothing convertible" in capsys.readouterr().err
+    said = capsys.readouterr().err
+    assert "no transcript" in said
+    # It must also say what it looked at: a bare refusal reads as the tool
+    # failing rather than the path not holding what was hoped for.
+    assert "1 file" in said and ".txt" in said
     assert not served
 
 

{transcript_viewer-0.7.2 → transcript_viewer-0.8.1}/tests/test_corpus.py

@@ -496,3 +496,36 @@ def test_pieces_unpacked_beside_a_download_are_used_as_they_are(tmp_path):
     entries = corpus.scan([tmp_path])
     assert len(entries) == 3
     assert all(str(beside) in e.path for e in entries)
+
+
+def test_a_path_that_converts_to_nothing_still_says_what_was_in_it(tmp_path):
+    """"Nothing convertible" alone reads as the tool failing, not the file."""
+    (tmp_path / "a").mkdir()
+    (tmp_path / "a" / "notes.md").write_text("# hello")
+    (tmp_path / "a" / "data.json").write_text('{"not": "a transcript"}')
+    (tmp_path / "a" / "run.py").write_text("print(1)")
+
+    said = corpus.describe_contents(tmp_path)
+
+    assert "3 files" in said
+    for ext in (".md", ".json", ".py"):
+        assert ext in said, f"{ext} should be named"
+
+
+def test_describing_a_single_file_names_it(tmp_path):
+    path = tmp_path / "one.jsonl"
+    path.write_text("{}\n")
+    said = corpus.describe_contents(path)
+    assert "one.jsonl" in said and "bytes" in said
+
+
+def test_describing_an_empty_directory_says_so(tmp_path):
+    assert corpus.describe_contents(tmp_path) == "it is empty"
+
+
+def test_describing_a_huge_tree_does_not_read_all_of_it(tmp_path):
+    """A 600 MB archive should not be walked twice to explain itself."""
+    for i in range(60):
+        (tmp_path / f"f{i}.bin").write_text("x")
+    said = corpus.describe_contents(tmp_path, limit=10)
+    assert "+" in said, "it should say the count is a floor, not a total"

{transcript_viewer-0.7.2 → transcript_viewer-0.8.1}/uv.lock

@@ -388,7 +388,7 @@ wheels = [
 
 [[package]]
 name = "transcript-viewer"
-version = "0.7.2"
+version = "0.8.1"
 source = { editable = "." }
 dependencies = [
     { name = "atif-make" },