search: implement whoosh search engine

* Updates README. * Also adds some cosmetic fixes. * Adds a CLI with commands to initialize the index and index all pages. Co-authored-by: Pascal Repond <[email protected]>
rero · Aug 24, 2023 · a937b1f · a937b1f
1 parent 89410af
commit a937b1f
Show file tree

Hide file tree

Showing 10 changed files with 212 additions and 67 deletions.
diff --git a/.gitignore b/.gitignore
@@ -128,4 +128,7 @@ dmypy.json
 *.mo
 
 # Macosx
-.DS_Store
+.DS_Store
+
+# Whoosh index
+/examples/index/
diff --git a/README.md b/README.md
@@ -8,16 +8,18 @@ Simple file based wiki for Flask.
 
 ### Requirements
 
-* Python >=3.6.2
+* Python >=3.8.0,<4.0.0
 * [Poetry](https://python-poetry.org/)
 
-### Install
+### Install dev environment
 
 - Clone the git repository
 - run `poetry install`
-- If you want to enable debug mode, run `export FLASK_ENV=development`
-- `cd examples`, then `poetry run flask run`
-- go to http://localhost:5000/wiki
+- `cd examples`, 
+- `poetry run flask flask_wiki init-index`
+- `poetry run flask flask_wiki index`
+- then `poetry run flask run --debug`
+- go to http://localhost:5000/help
 
 ## Configuration
 
@@ -36,8 +38,9 @@ Simple file based wiki for Flask.
 - WIKI_HOME = 'home'
 - WIKI_CURRENT_LANGUAGE = lambda: 'en'
 - WIKI_LANGUAGES = ['en']
-- WIKI_URL_PREFIX = '/wiki'
+- WIKI_URL_PREFIX = '/help'
 - WIKI_CONTENT_DIR = './data'
+- WIKI_INDEX_DIR = './index'
 - WIKI_UPLOAD_FOLDER = os.path.join(WIKI_CONTENT_DIR, 'files')
 - WIKI_ALLOWED_EXTENSIONS = {'png', 'jpg', 'jpeg', 'gif', 'svg'}
 - WIKI_MARKDOWN_EXTENSIONS = set(('codehilite', 'fenced_code'))

diff --git a/examples/app.py b/examples/app.py
@@ -49,7 +49,7 @@ def get_locale():
         app.config.from_mapping(test_config)
     Bootstrap4(app)
     Wiki(app)
-    babel = Babel(app, locale_selector=get_locale)
+    Babel(app, locale_selector=get_locale)
 
     @app.context_processor
     def inject_conf_var():

diff --git a/flask_wiki/api.py b/flask_wiki/api.py
@@ -10,15 +10,19 @@
 """Core classes."""
 
 import os
-import re
 from collections import OrderedDict
 from datetime import datetime
 from io import open
 from pathlib import Path
 
 import markdown
+from bs4 import BeautifulSoup
 from flask import abort, current_app, g
 from werkzeug.local import LocalProxy
+from whoosh import index, qparser
+from whoosh.analysis import LanguageAnalyzer
+from whoosh.fields import ID, TEXT, Schema
+from whoosh.writing import AsyncWriter
 
 from .markdown_ext import BootstrapExtension
 from .utils import clean_url, wikilink
@@ -44,7 +48,9 @@ def __init__(self, text):
 
         self.md = markdown.Markdown(extensions={
             BootstrapExtension(),
-            'codehilite', 'fenced_code', 'toc', 'meta', 'tables'
+            'codehilite',
+            'fenced_code',
+            'toc', 'meta', 'tables'
             }.union(markdown_ext))
 
         self.input = text
@@ -167,6 +173,19 @@ def render(self):
         self.modification_datetime = datetime.fromtimestamp(
             os.path.getmtime(self.path))
 
+    def index(self):
+        """Index page data for whoosh search engine."""
+        index_dir = index.open_dir(current_app.config.get('WIKI_INDEX_DIR'))
+        writer = AsyncWriter(index_dir)
+        writer.update_document(
+            url=self.url,
+            title=self.title,
+            body=self.raw_body,
+            tags=self.tags,
+            language=self.language
+        )
+        writer.commit()
+
     def save(self, update=True):
         """Save a page."""
         folder = os.path.dirname(self.path)
@@ -178,6 +197,7 @@ def save(self, update=True):
                 f.write(line)
             f.write(u'\n')
             f.write(self.body.replace(u'\r\n', u'\n'))
+        self.index()
         if update:
             self.load()
             self.render()
@@ -214,7 +234,6 @@ def title(self):
 
     @title.setter
     def title(self, value):
-        """."""
         self['title'] = value
 
     @property
@@ -227,9 +246,23 @@ def tags(self):
 
     @tags.setter
     def tags(self, value):
-        """."""
         self['tags'] = value
 
+    @property
+    def raw_body(self):
+        """Return raw text of the body.
+
+        Returns the raw text of the body without markdown or html markup,
+        used for indexing and search results display.
+        """
+        html = markdown.markdown(self.body)
+        html = BeautifulSoup(html, 'html.parser')
+        return html.get_text(separator=' ')
+
+    @raw_body.setter
+    def raw_body(self, value):
+        self['raw_body'] = value
+
     @property
     def language(self):
         """Return page language.
@@ -239,11 +272,11 @@ def language(self):
         """
         filename = Path(self.path).stem
         return filename.split('_')[-1] if '_' in filename\
-            else current_wiki.languages[0]
+            else list(current_wiki.languages.keys())[0]
 
 
 class WikiBase(object):
-    """."""
+    """Utility class for wiki management methods."""
 
     def __init__(self, root):
         """."""
@@ -309,9 +342,54 @@ def delete(self, url):
         if not self.exists(url):
             return False
         os.remove(path)
+        index_dir = index.open_dir(current_app.config.get('WIKI_INDEX_DIR'))
+        writer = AsyncWriter(index_dir)
+        writer.delete_by_term('path', path)
+        writer.commit()
         return True
 
-    def index(self):
+    def init_search_index(self):
+        """Create a new whoosh search index for the wiki."""
+        index_dir = current_app.config.get('WIKI_INDEX_DIR')
+        # initialize whoosh index schema
+        schema = Schema(
+            url=ID(stored=True, unique=True),
+            title=TEXT(stored=True, analyzer=LanguageAnalyzer("fr")),
+            tags=TEXT(stored=True),
+            body=TEXT(stored=True, analyzer=LanguageAnalyzer("fr")),
+            language=ID(stored=True)
+            )
+        if not os.path.exists(index_dir):
+            os.mkdir(index_dir)
+        index.create_in(index_dir, schema)
+
+    def search(self, query, ix, searcher):
+        """Search the whoosh index for a given query.
+
+        :param str query: the search query
+        :param whoosh.index ix: the whoosh index to use
+        :param whoosh.searcher searcher: an active whoosh searcher instance
+
+        :returns: a whoosh.results object instance
+        """
+        # parse the query to search all fields present in the schema
+        fields = ix.schema.names()
+        query_parser = qparser.MultifieldParser(
+            fields,
+            schema=ix.schema,
+            group=qparser.OrGroup
+            )
+        parsed_query = query_parser.parse(query)
+        # return a whoosh Results object to treat results
+        results = searcher.search(parsed_query)
+        # set highlights fragment size to 50 words
+        results.fragmenter.surround = 50
+        # set highlights separator for display
+        results.formatter.between = '<strong> [...] </strong>'
+        # return the modified Results object
+        return results
+
+    def list_pages(self):
         """Build up a list of all the available pages.
 
         :returns: a list of all the wiki pages
@@ -332,6 +410,11 @@ def index(self):
                     pages.append(page)
         return sorted(pages, key=lambda x: x.title.lower())
 
+    def index_all_pages(self):
+        """Index all the pages for the current wiki."""
+        for page in self.list_pages():
+            Page.index(page)
+
     def index_by(self, key):
         """Get an index based on the given key.
 
@@ -352,13 +435,13 @@ def index_by(self, key):
         return pages
 
     def get_by_title(self, title):
-        """."""
-        pages = self.index(attr='title')
+        """Get all page titles."""
+        pages = self.list_pages(attr='title')
         return pages.get(title)
 
     def get_tags(self):
-        """."""
-        pages = self.index()
+        """Get all tags."""
+        pages = self.list_pages()
         tags = {}
         for page in pages:
             pagetags = page.tags.split(',')
@@ -372,9 +455,9 @@ def get_tags(self):
                     tags[tag] = [page]
         return tags
 
-    def index_by_tag(self, tag):
-        """."""
-        pages = self.index()
+    def list_tagged_pages(self, tag):
+        """Get a list of all pages that have a tag."""
+        pages = self.list_pages()
         tagged = [page for page in pages if tag in page.tags]
         return sorted(tagged, key=lambda x: x.title.lower())
 
@@ -388,39 +471,6 @@ def languages(self):
         """."""
         return current_app.config.get('WIKI_LANGUAGES')
 
-    def search(self, term, ignore_case=True, attrs=None):
-        """."""
-        if attrs is None:
-            attrs = ['title', 'tags', 'body']
-        pages = self.index()
-
-        for page in pages:
-            page["score"] = 0
-
-        # When searching for "*", return ALL pages
-        if term == "*":
-            return pages
-
-        current_language_pages = [
-            p for p in pages if p.language == self.current_language]
-
-        # If no query term, return all current language pages
-        if not term:
-            return current_language_pages
-
-        regex = re.compile(
-            re.escape(term), re.IGNORECASE if ignore_case else 0)
-
-        matched = []
-        for page in current_language_pages:
-            for attr in attrs:
-                if found := re.findall(regex, getattr(page, attr)):
-                    page["score"] += len(found)
-                    if page not in matched:
-                        matched.append(page)
-        # Sort results by score
-        return sorted(matched, key=lambda x: x["score"], reverse=True)
-
 
 def get_wiki():
     """."""

diff --git a/flask_wiki/cli.py b/flask_wiki/cli.py
@@ -0,0 +1,35 @@
+# -*- coding: utf-8 -*-
+#
+# This file is part of Flask-Wiki
+# Copyright (C) 2023 RERO
+#
+# Flask-Wiki is free software; you can redistribute it and/or modify
+# it under the terms of the Revised BSD License; see LICENSE file for
+# more details.
+
+"""Click command-line interface for flask-wiki."""
+
+import click
+from flask.cli import with_appcontext
+
+from .api import get_wiki
+
+
+@click.group()
+def flask_wiki():
+    """Command-line interface for flask-wiki."""
+    pass
+
+
+@flask_wiki.command()
+@with_appcontext
+def init_index():
+    """Init whoosh search index."""
+    get_wiki().init_search_index()
+
+
+@flask_wiki.command()
+@with_appcontext
+def index():
+    """Index all wiki pages for whoosh search."""
+    get_wiki().index_all_pages()
diff --git a/flask_wiki/config.py b/flask_wiki/config.py
@@ -27,9 +27,10 @@
 WIKI_HOME = 'home'
 WIKI_CURRENT_LANGUAGE = lambda: 'en'
 WIKI_LANGUAGES = ['en']
-WIKI_URL_PREFIX = '/wiki'
+WIKI_URL_PREFIX = '/help'
 WIKI_CONTENT_DIR = './data'
 WIKI_UPLOAD_FOLDER = os.path.join(WIKI_CONTENT_DIR, 'files')
+WIKI_INDEX_DIR = './index'
 WIKI_ALLOWED_EXTENSIONS = {'png', 'jpg', 'jpeg', 'gif', 'svg'}
 
 """Markdown Extensions.

diff --git a/flask_wiki/templates/wiki/search.html b/flask_wiki/templates/wiki/search.html
@@ -19,20 +19,20 @@
         <header>
             <div class="pb-3">
                 {{ results | length }} {{ ngettext('result', 'results', results | length) }}
-                <a href="{{ url_for('wiki.search', q='*') }}" class="btn btn-sm btn-outline-primary ml-2">{{ _("All languages") }}</a>
                 </div>
         </header>
         <div>
             <ul class="list-group list-group-flush">
                 {%- for result in results -%}
                 <li class="list-group-item">
                     <div class="d-lg-flex justify-content-lg-between">
-                        <article class="mb-2">
+                        <article class="mb-2 col-10">
                             <h5 class="m-0">
                                 <a class="mr-2" href="{{ url_for('wiki.page', url=result.url) }}">
                                     {{ result.title }}
                                 </a>
                             </h5>
+                            <p>{{ result.highlights("body", top=3) | safe }}</p>
                             <ul class="m-0 p-0">
                                 <li class="badge badge-secondary">
                                     {{ result.language | upper }}
@@ -45,7 +45,7 @@ <h5 class="m-0">
                             </ul>
                         </article>
                         {% if can_edit_wiki %}
-                        <footer>
+                        <footer class="ml-3 col-4">
                             <button data-name="{{ result.title }}"
                                     data-link="{{ url_for('wiki.page', url=result.url) }}"
                                     class="copy-file-code btn btn-sm btn-outline-primary">