chat_sdk-telegram 0.6.0 → 0.8.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: a14fedc4aa2b77c45981da8147e152cb78770d131e1bbe234468655d0b171173
-  data.tar.gz: 3b35d0c60617dc93afa0b706b70e80f21d5672a4a00e0f7d0db374f49c6318c4
+  metadata.gz: 9c7469e27cabb92ba6ed37a198b0b654bd28e41d6be5c6a7d2e869aa422ca242
+  data.tar.gz: 6b6c66121c254a6ec4206130f6ff1001275ce84b202ec67995cfdafd2f38add8
 SHA512:
-  metadata.gz: b3da00035b7d571f92a05ec573212754999cf553f71608a6cf3473b9410f060951e3324a10ef520faa393dd16a17fe397d77794cdab8b175ab86909c0fc90501
-  data.tar.gz: fb63c59d2e3a750530ff211f392b35b3003b5afa8e7bb7f7c277149bebc6d32ca4cc69f85c473428be9419359b1162a6d7ed7843472343069cbeef503b6255f7
+  metadata.gz: 6f13191588083d918d717038d6946270de24fc34b578120009045c27c4bc64d1800f444846af8db9734e80edf8ae03c9eb26f2e33a149f422ba59f469eaf22b1
+  data.tar.gz: 61425a3e878f398cd1222d6827bee44a1e84ab035103c3606f734856fb8588b322c97945d274aaa23e59ab0f67602f09143a712c919174c4c9a56509eac2df2b

data/lib/chat_sdk/telegram/format_converter.rb

@@ -0,0 +1,124 @@
+# frozen_string_literal: true
+
+module ChatSDK
+  module Telegram
+    class FormatConverter < ChatSDK::Format::Converter
+      SPECIAL_CHARS = %w[_ * [ ] ( ) ~ \\ ` > # + - = | { } . !].freeze
+      ESCAPE_RE = Regexp.union(SPECIAL_CHARS).freeze
+
+      # Telegram MarkdownV2 → standard Markdown
+      def to_markdown(platform_text)
+        text = platform_text.to_s
+        return "" if text.empty?
+
+        # Convert Telegram user links: [text](tg://user?id=123) → @123
+        text = text.gsub(/\[([^\]]*)\]\(tg:\/\/user\?id=(\d+)\)/, '@\2')
+
+        # Strip underline markers: __text__ → text
+        # Use a loop to handle nested cases
+        text = text.gsub(/(?<!_)__(?!_)(.+?)(?<!_)__(?!_)/m, '\1')
+
+        # Strip spoiler markers: ||text|| → text
+        text = text.gsub(/\|\|(.+?)\|\|/m, '\1')
+
+        # Unescape all backslash-escaped characters
+        text.gsub(/\\(.)/, '\1')
+      end
+
+      # Standard Markdown → Telegram MarkdownV2
+      def from_markdown(markdown)
+        text = markdown.to_s
+        return "" if text.empty?
+
+        parts = split_code_segments(text)
+        parts.map.with_index { |part, i| i.odd? ? part : escape_special_chars(part) }.join
+      end
+
+      private
+
+      # Splits text into alternating [text, code, text, code, ...] segments.
+      # Odd-indexed segments are code blocks/inline code and are NOT escaped.
+      def split_code_segments(text)
+        text.split(/(```[\s\S]*?```|`[^`]+`)/)
+      end
+
+      # Escape Telegram MarkdownV2 special chars outside of markdown syntax.
+      # Preserves bold (**), italic (*), strikethrough (~~), links, etc.
+      def escape_special_chars(text)
+        result = +""
+        i = 0
+        chars = text.chars
+
+        while i < chars.length
+          char = chars[i]
+
+          # Bold: **text** — pass through
+          if char == "*" && chars[i + 1] == "*"
+            close = text.index("**", i + 2)
+            if close
+              inner = text[(i + 2)...close]
+              result << "**#{escape_special_chars(inner)}**"
+              i = close + 2
+              next
+            end
+          end
+
+          # Italic: *text* (single asterisk, not preceded/followed by another *)
+          if char == "*" && chars[i + 1] != "*"
+            close = find_single_asterisk_close(chars, i + 1)
+            if close
+              inner = chars[(i + 1)...close].join
+              result << "*#{escape_special_chars(inner)}*"
+              i = close + 1
+              next
+            end
+          end
+
+          # Strikethrough: ~~text~~
+          if char == "~" && chars[i + 1] == "~"
+            close = text.index("~~", i + 2)
+            if close
+              inner = text[(i + 2)...close]
+              result << "~~#{escape_special_chars(inner)}~~"
+              i = close + 2
+              next
+            end
+          end
+
+          # Links: [text](url) — pass through
+          if char == "["
+            link_match = text[i..].match(/\A\[([^\]]*)\]\(([^)]*)\)/)
+            if link_match
+              result << link_match[0]
+              i += link_match[0].length
+              next
+            end
+          end
+
+          # Escape special characters
+          result << if SPECIAL_CHARS.include?(char)
+            "\\#{char}"
+          else
+            char
+          end
+
+          i += 1
+        end
+
+        result
+      end
+
+      # Find closing single * that isn't part of **
+      def find_single_asterisk_close(chars, start)
+        i = start
+        while i < chars.length
+          if chars[i] == "*" && chars[i + 1] != "*" && (i == start || chars[i - 1] != "*")
+            return i
+          end
+          i += 1
+        end
+        nil
+      end
+    end
+  end
+end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: chat_sdk-telegram
 version: !ruby/object:Gem::Version
-  version: 0.6.0
+  version: 0.8.0
 platform: ruby
 authors:
 - Quentin Rousseau
@@ -48,6 +48,7 @@ files:
 - lib/chat_sdk/telegram/adapter.rb
 - lib/chat_sdk/telegram/api_client.rb
 - lib/chat_sdk/telegram/event_parser.rb
+- lib/chat_sdk/telegram/format_converter.rb
 - lib/chat_sdk/telegram/keyboard_renderer.rb
 homepage: https://github.com/rootlyhq/chat-sdk
 licenses: