[Wiki] Misc improvements, plus a beta search feature

3 files changed, 86 insertions, 5 deletions

diff --git a/pysite/database.py b/pysite/database.py
index 86c8685d..41fbd253 100644
--- a/pysite/database.py
+++ b/pysite/database.py
@@ -1,8 +1,10 @@
 # coding=utf-8
+import html
 import json
 import logging
 import os
 from typing import Any, Callable, Dict, Iterator, List, Optional, Union
+import re
 
 import rethinkdb
 from flask import abort
@@ -22,6 +24,9 @@ ALL_TABLES = {
     "wiki_revisions": "id"
 }
 
+STRIP_REGEX = re.compile(r"<[^<]+?>")
+WIKI_TABLE = "wiki"
+
 
 class RethinkDB:
 
@@ -55,6 +60,13 @@ class RethinkDB:
                     tables = ", ".join([f"{table} ({count} items)" for table, count in initialized.items()])
                     self.log.debug(f"Initialized the following tables: {tables}")
 
+                # Upgrade wiki articles
+                for article in self.pluck(WIKI_TABLE, "html", "text", "slug"):
+                    if "text" not in article:
+                        article["text"] = html.unescape(STRIP_REGEX.sub("", article["html"]).strip())
+
+                    self.insert(WIKI_TABLE, article, conflict="update")
+
     def create_tables(self) -> List[str]:
         """
         Creates whichever tables exist in the ALL_TABLES
@@ -496,9 +508,6 @@ class RethinkDB:
         """
         Map a function over every document in a table, with the possibility of modifying it
 
-        r.table('users').map(
-        lambda doc: doc.merge({'user_id': doc['id']}).without('id')).run(conn)
-
         As an example, you could do the following to rename the "id" field to "user_id" for all documents
         in the "users" table.
 
diff --git a/pysite/views/wiki/edit.py b/pysite/views/wiki/edit.py
index 2cd4181c..bb39ed2b 100644
--- a/pysite/views/wiki/edit.py
+++ b/pysite/views/wiki/edit.py
@@ -1,5 +1,7 @@
 import datetime
 import difflib
+import html
+import re
 
 import requests
 from flask import redirect, request, url_for
@@ -11,6 +13,8 @@ from pysite.decorators import csrf, require_roles
 from pysite.mixins import DBMixin
 from pysite.rst import render
 
+STRIP_REGEX = re.compile(r"<[^<]+?>")
+
 
 class EditView(RouteView, DBMixin):
     path = "/edit/<path:page>"  # "path" means that it accepts slashes
@@ -55,11 +59,12 @@ class EditView(RouteView, DBMixin):
     @csrf
     def post(self, page):
         rst = request.form.get("rst")
+        title = request.form["title"]
 
-        if not rst:
+        if not rst or not not rst.strip():
             raise BadRequest()
 
-        if not rst.strip():
+        if not title or not title.strip():
             raise BadRequest()
 
         rendered = render(rst)
@@ -69,6 +74,7 @@ class EditView(RouteView, DBMixin):
             "title": request.form["title"],
             "rst": rst,
             "html": rendered["html"],
+            "text": html.unescape(STRIP_REGEX.sub("", rendered["html"]).strip()),
             "headers": rendered["headers"]
         }
 
diff --git a/pysite/views/wiki/search.py b/pysite/views/wiki/search.py
new file mode 100644
index 00000000..1c920e15
--- /dev/null
+++ b/pysite/views/wiki/search.py
@@ -0,0 +1,66 @@
+import html
+import re
+
+from flask import redirect, request, url_for
+from werkzeug.exceptions import BadRequest
+
+from pysite.base_route import RouteView
+from pysite.decorators import csrf
+from pysite.mixins import DBMixin
+
+STRIP_REGEX = re.compile(r"<[^<]+?>")
+
+
+class SearchView(RouteView, DBMixin):
+    path = "/search"  # "path" means that it accepts slashes
+    name = "search"
+    table_name = "wiki"
+    revision_table_name = "wiki_revisions"
+
+    def get(self):
+        return self.render("wiki/search.html")
+
+    @csrf
+    def post(self):
+        given_query = request.form.get("query")
+
+        if not given_query or not given_query.strip():
+            raise BadRequest()
+
+        query = f"({re.escape(given_query)})"
+
+        pages = self.db.filter(
+            self.table_name,
+            lambda doc: doc["text"].match(query)
+        )
+
+        if len(pages) == 1:
+            slug = pages[0]["slug"]
+            return redirect(url_for("wiki.page", page=slug), code=303)
+
+        for obj in pages:
+            text = obj["text"]
+
+            matches = re.finditer(query, text)
+            snippets = []
+
+            for match in matches:
+                start = match.start() - 50
+
+                if start < 0:
+                    start = 0
+
+                end = match.end() + 50
+
+                if end > len(text):
+                    end = len(text)
+
+                match_text = text[start:end]
+                match_text = re.sub(query, r"<strong>\1</strong>", html.escape(match_text))
+
+                snippets.append(match_text.replace("\n", "<br />"))
+
+            obj["matches"] = snippets
+
+        pages = sorted(pages, key=lambda d: d["title"])
+        return self.render("wiki/search_results.html", pages=pages, query=given_query)