feat(orgfile): Parse links from document

Author: kristijanhusak

lua/orgmode/files/file.lua

@@ -5,6 +5,7 @@ local Headline = require('orgmode.files.headline')
 local ts = vim.treesitter
 local config = require('orgmode.config')
 local Block = require('orgmode.files.elements.block')
+local Link = require('orgmode.org.hyperlinks.link')
 
 ---@class OrgFileMetadata
 ---@field mtime number
@@ -598,6 +599,23 @@ function OrgFile:get_archive_file_location()
   return config:parse_archive_location(self.filename, archive_location)
 end
 
+memoize('get_links')
+---@return OrgLink[]
+function OrgFile:get_links()
+  self:parse(true)
+  local ts_query = ts_utils.get_query([[
+      (paragraph (expr) @links)
+      (drawer (contents (expr) @links))
+      (headline (item (expr)) @links)
+  ]])
+
+  local links = {}
+  for _, match in ts_query:iter_captures(self.root, self:_get_source()) do
+    vim.list_extend(links, Link.all_from_line(self:get_node_text(match)))
+  end
+  return links
+end
+
 ---@private
 ---@return string | nil
 function OrgFile:_get_directive(directive_name)

lua/orgmode/org/hyperlinks/link.lua

@@ -5,6 +5,8 @@ local Url = require('orgmode.org.hyperlinks.url')
 ---@field desc string | nil
 local Link = {}
 
+local pattern = '%[%[([^%]]+.-)%]%]'
+
 ---@param str string
 ---@return OrgLink
 function Link:new(str)
@@ -30,7 +32,6 @@ end
 function Link.at_pos(line, pos)
   local links = {}
   local found_link = nil
-  local pattern = '%[%[([^%]]+.-)%]%]'
   local position
   for link in line:gmatch(pattern) do
     local start_from = #links > 0 and links[#links].to or nil
@@ -49,4 +50,17 @@ function Link.at_pos(line, pos)
   return Link:new(found_link), position
 end
 
+function Link.all_from_line(line)
+  local links = {}
+  for link in line:gmatch(pattern) do
+    local start_from = #links > 0 and links[#links].to or nil
+    local from, to = line:find(pattern, start_from)
+    if from and to then
+      table.insert(links, Link:new(link))
+    end
+  end
+
+  return links
+end
+
 return Link

tests/plenary/files/file_spec.lua

@@ -738,4 +738,45 @@ describe('OrgFile', function()
       assert.are.same('112233', file:get_property('custom_id'))
     end)
   end)
+
+  describe('get_links', function()
+    it('should get all links from a file', function()
+      local file = load_file_sync({
+        'Top level [[https://google.com]]',
+        '',
+        '* TODO Headline link to file [[./some-file.org]]',
+        '  - list item link to [[https://duckduckgo.com][duck]]',
+        '',
+        '  :LOGBOOK:',
+        '  :TEST: And link in drawer [[https://github.com][github]]',
+        '  :END:',
+      })
+      local links = file:get_links()
+
+      assert.are.same(4, #links)
+      assert.are.same('https://google.com', links[1].url:to_string())
+      assert.is.Nil(links[1].desc)
+
+      assert.are.same('./some-file.org', links[2].url:to_string())
+      assert.is.Nil(links[2].desc)
+
+      assert.are.same('https://duckduckgo.com', links[3].url:to_string())
+      assert.are.same('duck', links[3].desc)
+
+      assert.are.same('https://github.com', links[4].url:to_string())
+      assert.are.same('github', links[4].desc)
+    end)
+
+    it('should get file level property', function()
+      local file = load_file_sync({
+        ':PROPERTIES:',
+        ':ID: 443355',
+        ':CUSTOM_ID: 112233',
+        ':END:',
+        '#+title: test',
+        '* TODO Headline 1',
+      })
+      assert.are.same('112233', file:get_property('custom_id'))
+    end)
+  end)
 end)