Fix pcre2 issue with ripgrep (closes #145) (#148)

* fix(tags): fix rg command if PCRE not available (closes #145)

Some linux distro do not compile ripgrep using the --pcre2 flag.
This commit checks if the --pcre2 flag is available and then tweaks the
ripgrep command accordingly.

At the moment, it will therefore not be possible to filter-out
hexademical color codes from #tags for people that do not have the
--pcre2 flag available. This will be documented in a later commit.

**Important**: the detection of the PCRE availability is done by trying
a simple rg search using _os.execute_. This function interface was
changed in Lua 5.2, which means it will break when Neovim will update to
Lua >5.1. The fix is already there in commented form.

Fixes #145

* refact: cleanup

* doc(tag): add note about rg --pcre2

* doc(tag): add note about rg --pcre2 in vim help

README.md

@@ -704,6 +704,16 @@ The only symbols allowed are:
 Numbers are allowed in tags, as long as a tag is not purely numeric. For example, #1984 is not a valid tag, but `#y1984`
 is.
 
+**Note**: When using the `#tag` notation, telekasten will try to differentiate
+an actual tag from an hexadecimal code (e.g. `#FF0000`) to prevent false
+matches. This is achieved by using the `--pcre2` flag of `ripgrep`. However,
+some linux distribution (mostly the ones based on Debian) do not compile ripgrep
+with this flag by default, making it impossible to use. If this is the case, the
+tag functions of telekasten will not be able to differentiate color codes from
+actual tags and will return everything. A workaround is to either use the
+`:tag:` notation or to recompile ripgrep locally with the appropriate flag (see
+issues # 115 and #145).
+
 **Note**: For proper highlighting, the `auto_set_filetype` option is set to `true` by default. This automatically
 switches the filetype of opened notes from `markdown` to `telekasten`, and also registers the syntax with telescope
 previewers for `.md` files.

doc/telekasten.txt

@@ -755,6 +755,17 @@ Note:~
   default. This automatically sets the to `telekasten`, and also registers the
   syntax with telescope previewers for `.md` files.
 
+Note:
+  When using the `#tag` notation, telekasten will try to differentiate an
+  actual tag from an hexadecimal code (e.g. `#FF0000`) to prevent false
+  matches. This is achieved by using the `--pcre2` flag of `ripgrep`. However,
+  some linux distribution (mostly the ones based on Debian) do not compile
+  ripgrep with this flag by default, making it impossible to use. If this is
+  the case, the tag functions of telekasten will not be able to differentiate
+  color codes from actual tags and will return everything. A workaround is to
+  either use the `:tag:` notation or to recompile ripgrep locally with the
+  appropriate flag (see issues # 115 and #145).
+
 --------------------------------------------------------------------------------
 Section 3.3: Templates                                    *telekasten.templates*
 

lua/taglinks/tagutils.lua

@@ -1,7 +1,9 @@
 local Job = require("plenary.job")
 
 local M = {}
-local hashtag_re =
+local hashtag_re = "(^|\\s|'|\")#[a-zA-ZÀ-ÿ]+[a-zA-ZÀ-ÿ0-9/\\-_]*"
+-- PCRE hashtag allows to remove the hex color codes from hastags
+local hashtag_re_pcre =
     "(^|\\s|'|\")((?!(#[a-fA-F0-9]{3})(\\W|$)|(#[a-fA-F0-9]{6})(\\W|$))#[a-zA-ZÀ-ÿ]+[a-zA-ZÀ-ÿ0-9/\\-_]*)"
 local colon_re = "(^|\\s):[a-zA-ZÀ-ÿ]+[a-zA-ZÀ-ÿ0-9/\\-_]*:"
 local yaml_re =
@@ -11,6 +13,7 @@ local function command_find_all_tags(opts)
     opts = opts or {}
     opts.cwd = opts.cwd or "."
     opts.templateDir = opts.templateDir or ""
+    opts.rg_pcre = opts.rg_pcre or false
 
     -- do not list tags in the template directory
     local globArg = ""
@@ -28,8 +31,20 @@ local function command_find_all_tags(opts)
         re = yaml_re
     end
 
-    return "rg",
+    local rg_args = {
-        {
+        "--vimgrep",
+        globArg,
+        "-o",
+        re,
+        "--",
+        opts.cwd,
+    }
+
+    -- PCRE engine allows to remove hex color codes from #hastags
+    if opts.rg_pcre then
+        re = hashtag_re_pcre
+
+        rg_args = {
             "--vimgrep",
             "--pcre2",
             globArg,
@@ -38,6 +53,9 @@ local function command_find_all_tags(opts)
             "--",
             opts.cwd,
         }
+    end
+
+    return "rg", rg_args
 end
 
 -- strips away leading ' or " , then trims whitespace

lua/telekasten.lua

@@ -2734,6 +2734,7 @@ local function FindAllTags(opts)
     opts.tag_notation = M.Cfg.tag_notation
     local templateDir = Path:new(M.Cfg.templates):make_relative(M.Cfg.home)
     opts.templateDir = templateDir
+    opts.rg_pcre = M.Cfg.rg_pcre
 
     if not global_dir_check() then
         return
@@ -2913,6 +2914,16 @@ local function Setup(cfg)
     M.Cfg.dailies = make_config_path_absolute(M.Cfg.dailies)
     M.Cfg.weeklies = make_config_path_absolute(M.Cfg.weeklies)
     M.Cfg.templates = make_config_path_absolute(M.Cfg.templates)
+
+    -- Check if ripgrep is compiled with --pcre
+    -- ! This will need to be fixed when neovim moves to lua >=5.2 by the following:
+    -- M.Cfg.rg_pcre = os.execute("echo 'hello' | rg --pcr2 hello &> /dev/null") or false
+
+    M.Cfg.rg_pcre = false
+    local has_pcre = os.execute("echo 'hello' | rg --pcre2 hello &> /dev/null")
+    if has_pcre == 0 then
+        M.Cfg.rg_pcre = true
+    end
 end
 
 M.find_notes = FindNotes