fix: remove empty strings and fix file exlcusion

src/helpers/excluded-files.ts

@@ -34,31 +34,33 @@ async function parseGitAttributes(content: string): Promise<GitAttributes[]> {
 }
 
 export async function getExcludedFiles(context: Context) {
-  const [gitIgnoreContent, gitAttributesContent] = await Promise.all([getFileContent(".gitignore", context), getFileContent(".gitattributes", context)]);
+  const [gitIgnoreContent, gitAttributesContent] = await Promise.all([getFileContent(context, ".gitignore"), getFileContent(context, ".gitattributes")]);
 
-  const gitAttributesLinguistGenerated = (await parseGitAttributes(gitAttributesContent))
-    .filter((v) => v.attributes["linguist-generated"])
-    .map((v) => v.pattern);
-  const gitIgnoreExcludedFiles = gitIgnoreContent.split("\n").filter((v) => !v.startsWith("#"));
+  const gitAttributesLinguistGenerated = gitAttributesContent
+    ? (await parseGitAttributes(gitAttributesContent)).filter((v) => v.attributes["linguist-generated"]).map((v) => v.pattern)
+    : [];
+  const gitIgnoreExcludedFiles = gitIgnoreContent ? gitIgnoreContent.split("\n").filter((v) => !v.startsWith("#")) : [];
   return [...gitAttributesLinguistGenerated, ...gitIgnoreExcludedFiles];
 }
 
-async function getFileContent(path: string, context: Context): Promise<string> {
+async function getFileContent(context: Context, path: string): Promise<string | null> {
   try {
     const response = await context.octokit.rest.repos.getContent({
       owner: context.payload.repository.owner.login,
       repo: context.payload.repository.name,
       path,
-      ref: context.payload.pull_request.head.sha,
+      ref: context.payload.pull_request.base.sha,
     });
 
     // GitHub API returns content as base64
     if ("content" in response.data && !Array.isArray(response.data)) {
       return Buffer.from(response.data.content, "base64").toString("utf-8");
     }
-    return "";
-  } catch {
-    // File doesn't exist
-    return "";
+    return null;
+  } catch (err) {
+    if (err instanceof Error && "status" in err && err.status === 404) {
+      return null;
+    }
+    throw context.logger.error(`Error fetching files to be excluded ${err}`);
   }
 }

src/helpers/pull-helpers/fetch-diff.ts

@@ -20,5 +20,5 @@ export async function fetchPullRequestDiff(context: Context, org: string, repo:
     return { diff: null };
   }
 
-  return await processPullRequestDiff(diff, tokenLimits, context.logger);
+  return await processPullRequestDiff(diff, tokenLimits, context);
 }

src/helpers/pull-helpers/pull-request-parsing.ts

@@ -40,14 +40,14 @@ function selectIncludedFiles(
   return includedFiles;
 }
 
-export async function processPullRequestDiff(diff: string, tokenLimits: TokenLimits, logger: Context["logger"]) {
-  const excludedFilePatterns = await getExcludedFiles();
+export async function processPullRequestDiff(diff: string, tokenLimits: TokenLimits, context: Context) {
+  const excludedFilePatterns = await getExcludedFiles(context);
   const sortedDiffs = await filterAndSortDiffs(diff, excludedFilePatterns);
 
-  const includedFiles = selectIncludedFiles(sortedDiffs, tokenLimits, logger);
+  const includedFiles = selectIncludedFiles(sortedDiffs, tokenLimits, context.logger);
 
   if (includedFiles.length === 0) {
-    logger.error(`Cannot include any files from diff without exceeding token limits.`);
+    context.logger.error(`Cannot include any files from diff without exceeding token limits.`);
     return { diff: null };
   }
 
@@ -66,29 +66,18 @@ async function encodeAsync(text: string, options: EncodeOptions): Promise<number
 
 // Helper to parse a diff into per-file diffs
 function parsePerFileDiffs(diff: string): { filename: string; diffContent: string }[] {
-  // regex to capture diff sections, including the last file
-  const diffPattern = /^diff --git a\/(.*?) b\/.*$/gm;
-  let match: RegExpExecArray | null;
-  const perFileDiffs = [];
-  let lastIndex = 0;
-
-  // iterate over each file in the diff
-  while ((match = diffPattern.exec(diff)) !== null) {
-    const filename = match[1];
-    const startIndex = match.index;
-
-    // if we have pushed a file into the array, "append" the diff content
-    if (perFileDiffs.length > 0) {
-      perFileDiffs[perFileDiffs.length - 1].diffContent = diff.substring(lastIndex, startIndex).trim();
-    }
-
-    perFileDiffs.push({ filename, diffContent: "" });
-    lastIndex = startIndex;
-  }
-  // append the last file's diff content
-  if (perFileDiffs.length > 0 && lastIndex < diff.length) {
-    perFileDiffs[perFileDiffs.length - 1].diffContent = diff.substring(lastIndex).trim();
-  }
-
-  return perFileDiffs;
+  // Split the diff string into chunks for each file
+  const fileDiffs = diff.split(/^diff --git /gm).filter((chunk) => chunk.trim() !== "");
+
+  return fileDiffs.map((chunk) => {
+    // Extract the filename and content from each chunk
+    const lines = chunk.split("\n");
+    const firstLine = lines.shift()?.trim() || "";
+    const [filename] = firstLine.split(" ").map((part) => part.replace(/^a\//, ""));
+
+    return {
+      filename,
+      diffContent: `diff --git ${firstLine}\n${lines.join("\n").trim()}`,
+    };
+  });
 }