convoviz 0.2.7__tar.gz → 0.2.9__tar.gz

{convoviz-0.2.7 → convoviz-0.2.9}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: convoviz
-Version: 0.2.7
+Version: 0.2.9
 Summary: Get analytics and visualizations on your ChatGPT data!
 Keywords: markdown,chatgpt,openai,visualization,analytics,json,export,data-analysis,obsidian
 Author: Mohamed Cheikh Sidiya
@@ -34,8 +34,7 @@ Convert your ChatGPT history into well-formatted Markdown files. Additionally, v
 ## Features
 
 - **YAML Headers**: Optional and included by default.
-- **Track message versions**: prompt/response edits included.
-- **Code Interpreter**: Environment code blocks and execution results.
+- **Inline Images**: Media attachments rendered directly in Markdown.
 - **Data Visualizations**: Word clouds, graphs, and more.
 - **Custom Instructions**: All your custom instructions in one JSON file.
 

{convoviz-0.2.7 → convoviz-0.2.9}/README.md

@@ -8,8 +8,7 @@ Convert your ChatGPT history into well-formatted Markdown files. Additionally, v
 ## Features
 
 - **YAML Headers**: Optional and included by default.
-- **Track message versions**: prompt/response edits included.
-- **Code Interpreter**: Environment code blocks and execution results.
+- **Inline Images**: Media attachments rendered directly in Markdown.
 - **Data Visualizations**: Word clouds, graphs, and more.
 - **Custom Instructions**: All your custom instructions in one JSON file.
 

{convoviz-0.2.7 → convoviz-0.2.9}/convoviz/analysis/wordcloud.py

@@ -139,7 +139,8 @@ def generate_wordclouds(
         text = group.plaintext("user", "assistant")
         if text.strip():
             img = generate_wordcloud(text, config)
-            img.save(output_dir / f"{week.strftime('%Y week %W')}.png", optimize=True)
+            # Format: 2024-W15.png (ISO week format)
+            img.save(output_dir / f"{week.strftime('%Y-W%W')}.png", optimize=True)
 
     for month, group in tqdm(
         month_groups.items(),
@@ -149,7 +150,8 @@ def generate_wordclouds(
         text = group.plaintext("user", "assistant")
         if text.strip():
             img = generate_wordcloud(text, config)
-            img.save(output_dir / f"{month.strftime('%Y %B')}.png", optimize=True)
+            # Format: 2024-03-March.png (consistent with folder naming)
+            img.save(output_dir / f"{month.strftime('%Y-%m-%B')}.png", optimize=True)
 
     for year, group in tqdm(
         year_groups.items(),
@@ -159,4 +161,5 @@ def generate_wordclouds(
         text = group.plaintext("user", "assistant")
         if text.strip():
             img = generate_wordcloud(text, config)
+            # Format: 2024.png
             img.save(output_dir / f"{year.strftime('%Y')}.png", optimize=True)

{convoviz-0.2.7 → convoviz-0.2.9}/convoviz/config.py

@@ -10,8 +10,8 @@ from pydantic import BaseModel, Field
 class FolderOrganization(str, Enum):
     """How to organize markdown output files in folders."""
 
-    FLAT = "flat"  # All files in one directory (default)
-    DATE = "date"  # Nested by year/month/week
+    FLAT = "flat"  # All files in one directory
+    DATE = "date"  # Nested by year/month (default)
 
 
 class AuthorHeaders(BaseModel):

{convoviz-0.2.7 → convoviz-0.2.9}/convoviz/io/assets.py

@@ -57,6 +57,22 @@ def resolve_asset_path(source_dir: Path, asset_id: str) -> Path | None:
         except Exception:
             pass
 
+    # 4. Try prefix match in user-* directories (new 2025 format)
+    try:
+        for user_dir in source_dir.glob("user-*"):
+            if user_dir.is_dir():
+                user_dir = user_dir.resolve()
+                candidates = list(user_dir.glob(f"{asset_id}*"))
+                files = [
+                    p.resolve()
+                    for p in candidates
+                    if p.is_file() and p.resolve().is_relative_to(user_dir)
+                ]
+                if files:
+                    return files[0]
+    except Exception:
+        pass
+
     return None
 
 

{convoviz-0.2.7 → convoviz-0.2.9}/convoviz/io/writers.py

@@ -2,6 +2,7 @@
 
 from os import utime as os_utime
 from pathlib import Path
+from urllib.parse import quote
 
 from orjson import OPT_INDENT_2, dumps
 from tqdm import tqdm
@@ -32,8 +33,8 @@ _MONTH_NAMES = [
 def get_date_folder_path(conversation: Conversation) -> Path:
     """Get the date-based folder path for a conversation.
 
-    Creates a nested structure: year/month/week
-    Example: 2024/03-March/Week-02/
+    Creates a nested structure: year/month
+    Example: 2024/03-March/
 
     Args:
         conversation: The conversation to get the path for
@@ -51,13 +52,7 @@ def get_date_folder_path(conversation: Conversation) -> Path:
     month_name = _MONTH_NAMES[month_num - 1]
     month = f"{month_num:02d}-{month_name}"
 
-    # Week folder: "Week-01" through "Week-05" (week of the month)
-    # Calculate which week of the month this date falls into
-    day = create_time.day
-    week_of_month = (day - 1) // 7 + 1
-    week = f"Week-{week_of_month:02d}"
-
-    return Path(year) / month / week
+    return Path(year) / month
 
 
 def save_conversation(
@@ -115,6 +110,62 @@ def save_conversation(
     return final_path
 
 
+def _generate_year_index(year_dir: Path, year: str) -> None:
+    """Generate a _index.md file for a year folder.
+
+    Args:
+        year_dir: Path to the year directory
+        year: The year string (e.g., "2024")
+    """
+    months = sorted(
+        [d.name for d in year_dir.iterdir() if d.is_dir()],
+        key=lambda m: int(m.split("-")[0]),
+    )
+
+    lines = [
+        f"# {year}",
+        "",
+        "## Months",
+        "",
+    ]
+
+    for month in months:
+        month_name = month.split("-", 1)[1] if "-" in month else month
+        lines.append(f"- [{month_name}]({month}/_index.md)")
+
+    index_path = year_dir / "_index.md"
+    index_path.write_text("\n".join(lines) + "\n", encoding="utf-8")
+
+
+def _generate_month_index(month_dir: Path, year: str, month: str) -> None:
+    """Generate a _index.md file for a month folder.
+
+    Args:
+        month_dir: Path to the month directory
+        year: The year string (e.g., "2024")
+        month: The month folder name (e.g., "03-March")
+    """
+    month_name = month.split("-", 1)[1] if "-" in month else month
+    files = sorted(
+        [f.name for f in month_dir.glob("*.md") if f.name != "_index.md"]
+    )
+
+    lines = [
+        f"# {month_name} {year}",
+        "",
+        "## Conversations",
+        "",
+    ]
+
+    for file in files:
+        title = file[:-3]  # Remove .md extension
+        encoded_file = quote(file)
+        lines.append(f"- [{title}]({encoded_file})")
+
+    index_path = month_dir / "_index.md"
+    index_path.write_text("\n".join(lines) + "\n", encoding="utf-8")
+
+
 def save_collection(
     collection: ConversationCollection,
     directory: Path,
@@ -151,6 +202,15 @@ def save_collection(
         filepath = target_dir / f"{sanitize(conv.title)}.md"
         save_conversation(conv, filepath, config, headers, source_path=collection.source_path)
 
+    # Generate index files for date organization
+    if folder_organization == FolderOrganization.DATE:
+        for year_dir in directory.iterdir():
+            if year_dir.is_dir() and year_dir.name.isdigit():
+                for month_dir in year_dir.iterdir():
+                    if month_dir.is_dir():
+                        _generate_month_index(month_dir, year_dir.name, month_dir.name)
+                _generate_year_index(year_dir, year_dir.name)
+
 
 def save_custom_instructions(
     collection: ConversationCollection,

{convoviz-0.2.7 → convoviz-0.2.9}/convoviz/models/message.py

@@ -28,6 +28,14 @@ class MessageContent(BaseModel):
     parts: list[Any] | None = None
     text: str | None = None
     result: str | None = None
+    # reasoning_recap content type
+    content: str | None = None
+    # thoughts content type (list of thought objects with summary/content/finished)
+    thoughts: list[Any] | None = None
+    # tether_quote content type
+    url: str | None = None
+    domain: str | None = None
+    title: str | None = None
 
 
 class MessageMetadata(BaseModel):
@@ -36,6 +44,7 @@ class MessageMetadata(BaseModel):
     model_slug: str | None = None
     invoked_plugin: dict[str, Any] | None = None
     is_user_system_message: bool | None = None
+    is_visually_hidden_from_conversation: bool | None = None
     user_context_message_data: dict[str, Any] | None = None
 
     model_config = ConfigDict(protected_namespaces=())
@@ -105,12 +114,48 @@ class Message(BaseModel):
             if self.content.parts:
                 return ""
 
+        # tether_quote: render as a blockquote with attribution (check before .text)
+        if self.content.content_type == "tether_quote":
+            return self._render_tether_quote()
         if self.content.text is not None:
             return self.content.text
         if self.content.result is not None:
             return self.content.result
+        # reasoning_recap content type uses 'content' field
+        if self.content.content is not None:
+            return self.content.content
+        # thoughts content type uses 'thoughts' field (list of thought objects)
+        if self.content.thoughts is not None:
+            return self._render_thoughts()
         raise MessageContentError(self.id)
 
+    def _render_thoughts(self) -> str:
+        """Render thoughts content (list of thought objects with summary/content)."""
+        if not self.content.thoughts:
+            return ""
+        summaries = []
+        for thought in self.content.thoughts:
+            if isinstance(thought, dict) and (summary := thought.get("summary")):
+                summaries.append(summary)
+        return "\n".join(summaries) if summaries else ""
+
+    def _render_tether_quote(self) -> str:
+        """Render tether_quote content as a blockquote."""
+        quote_text = self.content.text or ""
+        if not quote_text.strip():
+            return ""
+        # Format as blockquote with source
+        lines = [f"> {line}" for line in quote_text.strip().split("\n")]
+        blockquote = "\n".join(lines)
+        # Add attribution if we have title/domain/url
+        if self.content.title and self.content.url:
+            blockquote += f"\n> — [{self.content.title}]({self.content.url})"
+        elif self.content.domain and self.content.url:
+            blockquote += f"\n> — [{self.content.domain}]({self.content.url})"
+        elif self.content.url:
+            blockquote += f"\n> — <{self.content.url}>"
+        return blockquote
+
     @property
     def has_content(self) -> bool:
         """Check if the message has extractable content."""
@@ -132,26 +177,42 @@ class Message(BaseModel):
 
         Hidden if:
         1. It is empty (no text, no images).
-        2. It is an internal system message (not custom instructions).
-        3. It is a browser tool output (intermediate search steps).
+        2. Explicitly marked as visually hidden.
+        3. It is an internal system message (not custom instructions).
+        4. It is a browser tool output (intermediate search steps).
+        5. It is an assistant message targeting a tool (internal call).
+        6. It is code interpreter input (content_type="code").
+        7. It is browsing status (tether_browsing_display).
+        8. It is internal reasoning (thoughts, reasoning_recap from o1/o3).
         """
         if self.is_empty:
             return True
 
+        # Explicitly marked as hidden by OpenAI
+        if self.metadata.is_visually_hidden_from_conversation:
+            return True
+
         # Hide internal system messages
         if self.author.role == "system":
             # Only show if explicitly marked as user system message (Custom Instructions)
             return not self.metadata.is_user_system_message
 
-        # Hide browser tool outputs (usually intermediate search steps)
+        # Hide browser tool outputs (intermediate search steps)
         if self.author.role == "tool" and self.author.name == "browser":
             return True
 
-        # Hide assistant calls to browser tool (e.g. "search(...)") or code interpreter
-        if self.author.role == "assistant" and (
-            self.recipient == "browser" or self.content.content_type == "code"
-        ):
+        # Hide assistant messages targeting tools (e.g., search(...), code input)
+        # recipient="all" or None means it's for the user; anything else is internal
+        if self.author.role == "assistant" and self.recipient not in ("all", None):
             return True
 
-        # Hide browsing status messages
-        return self.content.content_type == "tether_browsing_display"
+        # Hide code interpreter input (content_type="code")
+        if self.author.role == "assistant" and self.content.content_type == "code":
+            return True
+
+        # Hide browsing status and internal reasoning steps (o1/o3 models)
+        return self.content.content_type in (
+            "tether_browsing_display",
+            "thoughts",
+            "reasoning_recap",
+        )

{convoviz-0.2.7 → convoviz-0.2.9}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "convoviz"
-version = "0.2.7"
+version = "0.2.9"
 description = "Get analytics and visualizations on your ChatGPT data!"
 license = "MIT"
 keywords = [