Merge pull request #1798 from kcaran/or-tag-search

.gitignore

@@ -27,6 +27,17 @@ phpmd.html
 phpdoc.xml
 
 # User plugin configuration
+plugins/*
+!addlink_toolbar
+!archiveorg
+!default_colors
+!demo_plugin
+!isso
+!piwik
+!playvideos
+!pubsubhubbub
+!qrcode
+!wallabag
 plugins/*/config.php
 plugins/default_colors/default_colors.css
 

application/bookmark/BookmarkFilter.php

@@ -323,7 +323,6 @@ public function filterTags($tags, bool $casesensitive = false, string $visibilit
             // we got an input string, split tags
             $inputTags = tags_str2array($inputTags, $tagsSeparator);
         }
-
         if (count($inputTags) === 0) {
             // no input tags
             return $this->noFilter($visibility);
@@ -341,7 +340,20 @@ public function filterTags($tags, bool $casesensitive = false, string $visibilit
         }
 
         // build regex from all tags
-        $re = '/^' . implode(array_map([$this, 'tag2regex'], $inputTags)) . '.*$/';
+        $re_and = implode(array_map([$this, 'tag2regex'], $inputTags));
+        $re = '/^' . $re_and;
+
+        $orTags = array_filter(array_map(function ($tag) {
+            return startsWith($tag, '~') ? substr($tag, 1) : null;
+        }, $inputTags));
+
+        $re_or = implode('|', array_map([$this, 'tag2matchterm'], $orTags));
+        if ($re_or) {
+            $re_or = '(' . $re_or . ')';
+            $re .= $this->term2match($re_or, false);
+        }
+
+        $re .= '.*$/';
         if (!$casesensitive) {
             // make regex case insensitive
             $re .= 'i';
@@ -460,33 +472,54 @@ public static function tagsStrToArray(string $tags, bool $casesensitive): array
     /**
      * generate a regex fragment out of a tag
      *
-     * @param string $tag to to generate regexs from. may start with '-' to negate, contain '*' as wildcard
+     * @param string $tag to generate regexs from. may start with '-'
+     * to negate, contain '*' as wildcard. Tags starting with '~' are
+     * treated separately as an 'OR' clause.
      *
      * @return string generated regex fragment
      */
     protected function tag2regex(string $tag): string
     {
         $tagsSeparator = $this->conf->get('general.tags_separator', ' ');
-        $len = strlen($tag);
+        if (!$tag || $tag === "-" || $tag === "*" || $tag[0] === "~") {
-        if (!$len || $tag === "-" || $tag === "*") {
             // nothing to search, return empty regex
             return '';
         }
+        $negate = false;
+        if ($tag[0] === "+" && $tag[1]) {
+            $tag = substr($tag, 1); // use offset to start after '+' character
+        }
         if ($tag[0] === "-") {
             // query is negated
-            $i = 1; // use offset to start after '-' character
+            $tag = substr($tag, 1); // use offset to start after '-' character
-            $regex = '(?!'; // create negative lookahead
+            $negate = true;
-        } else {
-            $i = 0; // start at first character
-            $regex = '(?='; // use positive lookahead
         }
-        // before tag may only be the separator or the beginning
+        $term = $this->tag2matchterm($tag);
-        $regex .= '.*(?:^|' . $tagsSeparator . ')';
+
+        return $this->term2match($term, $negate);
+    }
+
+    /**
+     * generate a regex match term fragment out of a tag
+     *
+     * @param string $tag to to generate regexs from. This function
+     * assumes any leading flags ('-', '~') have been stripped. The
+     * wildcard flag '*' is expanded by this function and any other
+     * regex characters are escaped.
+     *
+     * @return string generated regex match term fragment
+     */
+    protected function tag2matchterm(string $tag): string
+    {
+        $tagsSeparator = $this->conf->get('general.tags_separator', ' ');
+        $len = strlen($tag);
+        $term = '';
         // iterate over string, separating it into placeholder and content
+        $i = 0; // start at first character
         for (; $i < $len; $i++) {
             if ($tag[$i] === '*') {
                 // placeholder found
-                $regex .= '[^' . $tagsSeparator . ']*?';
+                $term .= '[^' . $tagsSeparator . ']*?';
             } else {
                 // regular characters
                 $offset = strpos($tag, '*', $i);
@@ -497,11 +530,33 @@ protected function tag2regex(string $tag): string
                 // subtract one, as we want to get before the placeholder or end of string
                 $offset -= 1;
                 // we got a tag name that we want to search for. escape any regex characters to prevent conflicts.
-                $regex .= preg_quote(substr($tag, $i, $offset - $i + 1), '/');
+                $term .= preg_quote(substr($tag, $i, $offset - $i + 1), '/');
                 // move $i on
                 $i = $offset;
             }
         }
+
+        return $term;
+    }
+
+    /**
+     * generate a regex fragment out of a match term
+     *
+     * @param string $term is the match term already generated by tag2matchterm
+     * @param bool $negate if true create a negative lookahead
+     *
+     * @return string generated regex fragment
+     */
+    protected function term2match(string $term, bool $negate): string
+    {
+        $tagsSeparator = $this->conf->get('general.tags_separator', ' ');
+        $regex = $negate ? '(?!' : '(?='; // use negative or positive lookahead
+
+        // before tag may only be the separator or the beginning
+        $regex .= '.*(?:^|' . $tagsSeparator . ')';
+
+        $regex .= $term;
+
         // after the tag may only be the separator or the end
         $regex .= '(?:$|' . $tagsSeparator . '))';
         return $regex;

assets/default/js/base.js

@@ -45,11 +45,22 @@ function refreshToken(basePath, callback) {
 function createAwesompleteInstance(element, separator, tags = []) {
   const awesome = new Awesomplete(Awesomplete.$(element));
 
-  // Tags are separated by separator
+  // Tags are separated by separator. Ignore leading search flags
-  awesome.filter = (text, input) => Awesomplete.FILTER_CONTAINS(text, input.match(new RegExp(`[^${separator}]*$`))[0]);
+  awesome.filter = (text, input) => {
+    let filterFunc = Awesomplete.FILTER_CONTAINS;
+    let term = input.match(new RegExp(`[^${separator}]*$`))[0];
+    const termFlagged = term.replace(/^[-~+]/, '');
+    if (term !== termFlagged) {
+      term = termFlagged;
+      filterFunc = Awesomplete.FILTER_STARTSWITH;
+    }
+
+    return filterFunc(text, term);
+  };
+
   // Insert new selected tag in the input
   awesome.replace = (text) => {
-    const before = awesome.input.value.match(new RegExp(`^.+${separator}+|`))[0];
+    const before = awesome.input.value.match(new RegExp(`^(.+${separator}+)?[-~+]?|`))[0];
     awesome.input.value = `${before}${text}${separator}`;
   };
   // Highlight found items

doc/md/Usage.md

@@ -49,19 +49,21 @@ Shaarli can be used as a minimal blog, notepad, pastebin...: While adding or edi
 ### Search
 
 - **Plain text search:** Use `Search text` to search in all fields of all Shaares (Title, URL, Description...). Use double-quotes (example `"exact search"`) to search for the exact expression.
-- **Tags search:** `Filter by tags` allow only displaying Shaares tagged with one or multiple tags (use space to separate tags).
+- **Tags search:** `Filter by tags` allow only displaying Shaares tagged with one or multiple tags (use space to separate tags). A plus sign `+` is optional and will restrict suggested tags to only those starting with the string (example: `pr` will hint `apron` and `printer` but `+pr` will only hint printer).
 - **Hidden tags:** tags starting with a dot `.` (example `.secret`) are private. They can only be seen and searched when logged in.
 - **Exclude text/tags:** Use the `-` operator before a word or tag to exclude Shaares matching this word from search results (`NOT` operator).
+- **Optional tags:** Use the `~` operator before multiple tags to search for any one of them (`OR` operator).
+- **Wildcard tag search:** An asterisk (`*`) can be used as a wildcard and will match any number of characters. Wildcards can appear in the middle of a search term or at the end (example: pro\*in\* will match programming and protein).
 - **Untagged links:** Shaares without tags can be searched by clicking the `untagged` toggle button top left of the Shaares list (only when logged in).
 
-Both exclude patterns and exact searches can be combined with normal searches (example `"exact search" term otherterm -notthis "very exact" stuff -notagain`). Only AND (and NOT) search is currrently supported.
+Both exclude patterns and exact searches can be combined with normal searches (example `"exact search" term otherterm -notthis "very exact" stuff -notagain`). Only AND (and NOT) text search is currently supported.
 
-Active search terms are displayed on top of the link list. To remove terms/tags from the curent search, click the `x` next to any of them, or simply clear text/tag search fields.
+Active search terms are displayed on top of the link list. To remove terms/tags from the current search, click the `x` next to any of them, or simply clear text/tag search fields.
 
 
 ### Tag cloud
 
-The `Tag cloud` page diplays a "cloud" or list view of all tags in your Shaarli (most frequently used tags are displayed with a bigger font size)
+The `Tag cloud` page displays a "cloud" or list view of all tags in your Shaarli (most frequently used tags are displayed with a bigger font size)
 
 
 - **Tags list:** click on `Most used` or `Alphabetical` to display tags as a list. You can also edit/delete tags for this page.
@@ -76,7 +78,7 @@ RSS/ATOM feeds feeds are available (in ATOM with `/feed/atom` and RSS with `/fee
 
 - **Filtering RSS feeds:** RSS feeds and picture wall can also be restricted to only return items matching a text/tag search. For example, search for `photography` (text or tags) in Shaarli, then click the `RSS Feed` button. A feed with only matching results is displayed.
 - Add the `&nb` parameter in feed URLs to specify the number of Shaares you want in a feed (default if not specified: `50`). The keyword `all` is available if you want everything.
-- Add the `&permalinks` parameter in feed URLs to point permalinks to the corresponding shaarly entry/link instead of the direct, Shaare URL attribute
+- Add the `&permalinks` parameter in feed URLs to point permalinks to the corresponding Shaarli entry/link instead of the direct, Shaare URL attribute
 
 ![](images/rss-filter-1.png) ![](images/rss-filter-2.png)
 

tests/bookmark/BookmarkFilterTest.php

@@ -124,6 +124,11 @@ public function testFilterOneTag()
             count(self::$linkFilter->filter(BookmarkFilter::$FILTER_TAG, 'web', false))
         );
 
+        $this->assertEquals(
+            4,
+            count(self::$linkFilter->filter(BookmarkFilter::$FILTER_TAG, '+web', false))
+        );
+
         $this->assertEquals(
             4,
             count(self::$linkFilter->filter(BookmarkFilter::$FILTER_TAG, 'web', false, 'all'))
@@ -451,6 +456,37 @@ public function testFilterCrossedSearch()
         );
     }
 
+    /**
+     * Tag search with OR optional tags.
+     */
+    public function testTagFilterOr()
+    {
+        $this->assertEquals(
+            5,
+            count(self::$linkFilter->filter(BookmarkFilter::$FILTER_TAG, '~cartoon ~web'))
+        );
+
+        $this->assertEquals(
+            6,
+            count(self::$linkFilter->filter(BookmarkFilter::$FILTER_TAG, '~c*t*n ~st*'))
+        );
+
+        $this->assertEquals(
+            2,
+            count(self::$linkFilter->filter(BookmarkFilter::$FILTER_TAG, '~cartoon ~web dev'))
+        );
+
+        $this->assertEquals(
+            2,
+            count(self::$linkFilter->filter(BookmarkFilter::$FILTER_TAG, '~cartoon ~web +dev'))
+        );
+
+        $this->assertEquals(
+            4,
+            count(self::$linkFilter->filter(BookmarkFilter::$FILTER_TAG, '~cartoon ~web -samba'))
+        );
+    }
+
     /**
      * Filter bookmarks by #hashtag.
      */