hermitcrm 0.5.0__py3-none-any.whl

hermitcrm/__init__.py

@@ -0,0 +1,17 @@
+# Copyright 2026 Gijs Bos
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""Hermit CRM: your CRM is a folder of Markdown files in git."""
+
+__version__ = "0.5.0"

hermitcrm/accesslog.py

@@ -0,0 +1,57 @@
+# Copyright 2026 Gijs Bos
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""The serve log, without what a capture read off a page.
+
+The bookmarklet hands the running app what it read on a profile -- a name, a
+headline, an employer -- in the query string of a local address, because a
+new tab is the one thing a bookmarklet may open from any page. uvicorn writes
+every address it answers to the serve log, so without this the log would keep
+a copy of each person captured, outside the data folder and its history.
+
+The request line stays (path, status, time); only the query goes, and only on
+the capture paths. Everything else is logged exactly as uvicorn logs it.
+"""
+
+from __future__ import annotations
+
+import copy
+import logging
+
+from uvicorn.config import LOGGING_CONFIG
+
+# The page's own words arrive here; /capture/new is the old bookmarklet's path.
+QUIET_PATHS = ("/extension/new", "/capture/new")
+
+
+class QuietCaptures(logging.Filter):
+    """Drop the query of a capture request from uvicorn's access line."""
+
+    def filter(self, record: logging.LogRecord) -> bool:
+        args = record.args
+        # uvicorn: (client, method, path with query, http version, status)
+        if isinstance(args, tuple) and len(args) == 5 and isinstance(args[2], str):
+            path, _, query = args[2].partition("?")
+            if query and path in QUIET_PATHS:
+                record.args = (*args[:2], path + "?…", *args[3:])
+        return True
+
+
+def log_config() -> dict:
+    """uvicorn's own logging setup, with QuietCaptures on its access log."""
+    config = copy.deepcopy(LOGGING_CONFIG)
+    config.setdefault("filters", {})["quiet_captures"] = {"()": QuietCaptures}
+    access = config["loggers"].setdefault("uvicorn.access", {})
+    access["filters"] = [*access.get("filters", []), "quiet_captures"]
+    return config

hermitcrm/ask.py

@@ -0,0 +1,194 @@
+# Copyright 2026 Gijs Bos
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""Ask the Hermit: answer a question about the CRM with the configured AI CLI.
+
+Two steps, cheapest first. The page the question was asked on is rendered to
+plain text and the model answers from that text alone, without tools. Only
+when it says the page is not enough does a second run start inside the data
+folder with read-only tools (files plus `hermitcrm show/digest/report`), where
+the folder's CLAUDE.md tells an agent how to read the CRM cheaply. The model
+tier (medium by default, strong for "Retry with ...") comes from the Enricher.
+"""
+
+from __future__ import annotations
+
+import sys
+from dataclasses import dataclass, field
+from html.parser import HTMLParser
+from pathlib import Path
+
+from .enrich import EnrichError, Enricher, model_label
+
+PAGE_LIMIT = 40_000  # characters of page text sent with the first step
+
+# Tags whose content is chrome or form plumbing, not what the page says.
+SKIP_TAGS = {"script", "style", "nav", "header", "footer", "button", "svg", "template",
+             "noscript"}
+BLOCK_TAGS = {"p", "div", "section", "article", "main", "table", "tr", "li", "ul",
+              "ol", "h1", "h2", "h3", "h4", "h5", "h6", "pre", "details", "summary",
+              "br", "form", "label", "dl", "dt", "dd", "blockquote"}
+VOID_TAGS = {"br", "img", "input", "meta", "link", "hr", "col", "source", "wbr"}
+
+# Claude Code --allowedTools for the whole-CRM step: read-only.
+CRM_TOOLS = ("Read,Grep,Glob,Bash(hermitcrm show:*),Bash(hermitcrm digest:*),"
+             "Bash(hermitcrm report:*),Bash(hermitcrm check:*)")
+
+PAGE_SCHEMA = {
+    "type": "object",
+    "properties": {
+        "answerable": {"type": "boolean",
+                       "description": "true only if the page text is enough to answer"},
+        "answer": {"type": "string",
+                   "description": "the answer in Markdown; empty when not answerable"},
+    },
+    "required": ["answerable", "answer"],
+    "additionalProperties": False,
+}
+CRM_SCHEMA = {
+    "type": "object",
+    "properties": {
+        "answer": {"type": "string", "description": "the answer in Markdown"},
+        "sources": {"type": "array", "items": {"type": "string"},
+                    "description": "files or commands the answer relies on"},
+    },
+    "required": ["answer", "sources"],
+    "additionalProperties": False,
+}
+
+
+class _TextExtractor(HTMLParser):
+    def __init__(self) -> None:
+        super().__init__(convert_charrefs=True)
+        self.parts: list[str] = []
+        self.skip = 0
+        self.select = ""        # name of the <select> we are in, if any
+        self.selected = False   # inside its selected <option>: that value is page data
+
+    def handle_starttag(self, tag, attrs):
+        if tag == "select":
+            self.select = dict(attrs).get("name") or "choice"
+        elif tag == "option":
+            self.selected = self.select != "" and "selected" in dict(attrs)
+        elif tag in SKIP_TAGS and tag not in VOID_TAGS:
+            self.skip += 1
+        elif tag in BLOCK_TAGS:
+            self.parts.append("\n")
+        elif tag in ("td", "th"):
+            self.parts.append(" | ")
+        elif tag == "input" and not self.skip:
+            values = dict(attrs)
+            if values.get("type") not in ("hidden", "submit", "checkbox", "radio") \
+                    and values.get("value"):
+                self.parts.append(f" [{values.get('name', 'field')}: {values['value']}] ")
+
+    def handle_endtag(self, tag):
+        if tag == "select":
+            self.select, self.selected = "", False
+        elif tag == "option":
+            self.selected = False
+        elif tag in SKIP_TAGS and tag not in VOID_TAGS:
+            self.skip = max(0, self.skip - 1)
+        elif tag in BLOCK_TAGS:
+            self.parts.append("\n")
+
+    def handle_data(self, data):
+        if self.skip:
+            return
+        if self.select:  # of a dropdown only the chosen value is content
+            if self.selected and data.strip():
+                self.parts.append(f" [{self.select}: {data.strip()}] ")
+            return
+        self.parts.append(data)
+
+
+def page_text(html: str, limit: int = PAGE_LIMIT) -> str:
+    """The visible content of a rendered page as compact text, capped at limit."""
+    parser = _TextExtractor()
+    parser.feed(html or "")
+    parser.close()
+    lines = [" ".join(line.split()) for line in "".join(parser.parts).splitlines()]
+    text = "\n".join(line for line in lines if line.strip(" |"))
+    if len(text) > limit:
+        text = text[:limit] + "\n[page text cut off here]"
+    return text
+
+
+def page_prompt(question: str, page: str, text: str) -> str:
+    return (
+        "You are Hermit, the assistant inside Hermit CRM, a personal sales CRM. "
+        f"The user is looking at the page {page} and asks a question about it. "
+        "Answer using only the page content below. If the page does not contain "
+        "enough to answer reliably, set answerable to false and leave answer empty; "
+        "a search of the whole CRM follows. Be concise and concrete; use Markdown "
+        "lists or tables when they help.\n\n"
+        f"Question: {question}\n\n--- page content ---\n{text}\n--- end of page ---\n"
+    )
+
+
+def crm_prompt(question: str, page: str, text: str) -> str:
+    context = text[:4000]
+    return (
+        "You are Hermit, the assistant inside Hermit CRM. The current directory is "
+        "the CRM data folder; CLAUDE.md in it explains the layout and the cheapest "
+        "way to read it (PIPELINE.md first, `hermitcrm show <slug>`, `hermitcrm "
+        "digest`, `hermitcrm report --md`). Answer the question from that data. "
+        "Read only: do not create, edit or delete anything. If the data does not "
+        "answer the question, say so plainly. Be concise and concrete.\n\n"
+        f"The user asked this while on the page {page} (start of that page below, "
+        "for context).\n\n"
+        f"Question: {question}\n\n--- page ---\n{context}\n--- end ---\n"
+    )
+
+
+@dataclass
+class Answer:
+    question: str
+    text: str
+    scope: str                       # "page" or "crm"
+    model: str
+    tier: str
+    sources: list[str] = field(default_factory=list)
+
+    @property
+    def model_label(self) -> str:
+        return model_label(self.model)
+
+
+def ask(enricher: Enricher, question: str, page: str, html: str, data_dir: Path | str,
+        scope: str = "auto", timeout: float | None = None) -> Answer:
+    """Answer from the page first (scope auto), or go straight to the CRM (scope crm).
+
+    Raises EnrichError when no AI CLI is available or a run fails.
+    """
+    question = " ".join((question or "").split())
+    if not question:
+        raise EnrichError("Type a question first.")
+    if not enricher.available:
+        raise EnrichError(enricher.unavailable_reason())
+    if timeout:
+        enricher.timeout = float(timeout)
+    text = page_text(html)
+    if scope != "crm":
+        data = enricher.runner(page_prompt(question, page, text), PAGE_SCHEMA, tools="")
+        answer = str(data.get("answer") or "").strip()
+        if data.get("answerable") and answer:
+            return Answer(question, answer, "page", enricher.model, enricher.tier)
+    data = enricher.runner(crm_prompt(question, page, text), CRM_SCHEMA, tools=CRM_TOOLS,
+                           cwd=str(data_dir), extra_path=str(Path(sys.executable).parent))
+    answer = str(data.get("answer") or "").strip()
+    if not answer:
+        raise EnrichError("The AI returned an empty answer.")
+    sources = [str(s) for s in (data.get("sources") or []) if s]
+    return Answer(question, answer, "crm", enricher.model, enricher.tier, sources)