use dev markitdown

- contains fix for `pre` microsoft/markitdown#322
- DocumentConverterResult moved

Author: t-kalinowski

NAMESPACE

@@ -1,6 +1,5 @@
 # Generated by roxygen2: do not edit by hand
 
-S3method(reticulate::py_to_r,markitdown._markitdown.DocumentConverterResult)
 export(RagnarStore)
 export(embed_ollama)
 export(embed_openai)

R/ragnar-package.R

@@ -4,36 +4,42 @@
 
 .globals <- new.env(parent = emptyenv())
 
-#' @exportS3Method reticulate::py_to_r
-py_to_r.markitdown._markitdown.DocumentConverterResult <- function(x) {
+# ' @exportS3Method reticulate::py_to_r
+py_to_r.markitdown.DocumentConverterResult <- function(x) {
   c(title = x$title, text = x$text_content)
 }
 
 
 .onLoad <- function(libname, pkgname) {
   Sys.setenv(RETICULATE_PYTHON = "managed")
-  reticulate::py_require("markitdown")
-  reticulate:::py_register_load_hook("markitdown", function() {
+  reticulate::py_require(c("markitdown" = "git+https://github.com/microsoft/markitdown.git@main#subdirectory=packages/markitdown"))
 
-    `_CustomMarkdownify` <- reticulate::import("markitdown._markitdown")$`_CustomMarkdownify`
-    og_convert_a <- `_CustomMarkdownify`$convert_a
-    `_CustomMarkdownify`$convert_a <- function(self, el, text, convert_as_inline) {
-      # patch to prevent generating links in the markdown if
-      # - if the link is in a pre-formatted code block
-      # - if the link is to rdrr.io/r (autogenerated by pkgdown)
-      tryCatch({
-        if (el$get("href", "") |> startsWith("https://rdrr.io/r")) {
-          return(text)
-        }
-
-        if (!is.null(el$find_parent("pre"))) {
-          return(text)
-        }
-      }, error = warning)
-
-      og_convert_a(self, el, text, convert_as_inline)
-    }
+  reticulate:::py_register_load_hook("markitdown", function() {
 
+    registerS3method(
+      "py_to_r",
+      nameOfClass(reticulate::import("markitdown")$DocumentConverterResult),
+      py_to_r.markitdown.DocumentConverterResult,
+      environment(reticulate::py_to_r)
+    )
+
+    # `_CustomMarkdownify` <- reticulate::import("markitdown._markitdown")$`_CustomMarkdownify`
+    # og_convert_a <- `_CustomMarkdownify`$convert_a
+    # `_CustomMarkdownify`$convert_a <- function(self, el, text, convert_as_inline) {
+    #   # patch to prevent generating links in the markdown if
+    #   # - if the link is in a pre-formatted code block
+    #   # - if the link is to rdrr.io/r (autogenerated by pkgdown)
+    #   tryCatch({
+    #     if (el$get("href", "") |> startsWith("https://rdrr.io/r")) {
+    #       return(text)
+    #     }
+    #
+    #     if (!is.null(el$find_parent("pre"))) {
+    #       return(text)
+    #     }
+    #   }, error = warning)
+    #
+    #   og_convert_a(self, el, text, convert_as_inline)
+    # }
   })
 }
-