Patedam/Juliet
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Regenerated doc + final submit for now

Browse Source
This commit is contained in:
Patedam
2026-07-22 17:37:01 -04:00
parent c436639ddd
commit c0d40ef3e4
101 changed files with 869 additions and 781 deletions
Download Patch File Download Diff File Expand all files Collapse all files
Romeo/src/AiGenerator.cpp
+91 -13
View File
@@ -170,6 +170,71 @@ namespace Romeo
return true;
}
static Juliet::String CleanAndFormatDescription(Juliet::Arena* arena, const Juliet::String& rawResponse, bool enableCoT)
{
Assert(arena != nullptr);
if (rawResponse.Size == 0)
{
return {};
}
const char* pStr = CStr(rawResponse);
size_t len = rawResponse.Size;
const char* thinkStart = strstr(pStr, "<think>");
const char* thinkEnd = thinkStart ? strstr(thinkStart, "</think>") : nullptr;
Juliet::String thinkContent = {};
Juliet::String mainContent = {};
if (thinkStart && thinkEnd && thinkEnd > thinkStart)
{
const char* thinkTextStart = thinkStart + 7;
size_t thinkLen = static_cast<size_t>(thinkEnd - thinkTextStart);
thinkContent = { const_cast<char*>(thinkTextStart), thinkLen };
const char* afterThink = thinkEnd + 8;
size_t remainingLen = len - static_cast<size_t>(afterThink - pStr);
mainContent = { const_cast<char*>(afterThink), remainingLen };
}
else
{
mainContent = rawResponse;
}
// Clean mainContent (remove titles, markdown headers, leading quotes/fences)
const char* mPtr = CStr(mainContent);
size_t mLen = mainContent.Size;
while (mLen > 0 && (*mPtr == ' ' || *mPtr == '\t' || *mPtr == '\r' || *mPtr == '\n' || *mPtr == '#' || *mPtr == '`'))
{
mPtr++;
mLen--;
}
while (mLen > 0 && (mPtr[mLen - 1] == ' ' || mPtr[mLen - 1] == '\t' || mPtr[mLen - 1] == '\r' || mPtr[mLen - 1] == '\n' || mPtr[mLen - 1] == '`'))
{
mLen--;
}
if (enableCoT && thinkContent.Size > 0)
{
size_t resultCap = thinkContent.Size + mLen + 256;
char* resultBuf = Juliet::ArenaPushArray<char>(arena, resultCap JULIET_DEBUG_PARAM("CoTFormattedDesc"));
int written = snprintf(resultBuf, resultCap,
"<details>\n<summary>Chain of Thought</summary>\n\n%.*s\n\n</details>\n\n%.*s",
static_cast<int>(thinkContent.Size), CStr(thinkContent),
static_cast<int>(mLen), mPtr
);
return { resultBuf, static_cast<size_t>(written > 0 ? written : 0) };
}
char* resultBuf = Juliet::ArenaPushArray<char>(arena, mLen + 1 JULIET_DEBUG_PARAM("CleanedDesc"));
memcpy(resultBuf, mPtr, mLen);
resultBuf[mLen] = '\0';
return { resultBuf, mLen };
}
bool AI_GenerateDescription(
Juliet::Arena* arena,
FileEntry& entry
@@ -186,8 +251,23 @@ namespace Romeo
return true;
}
// Formulate the prompt
const char* sysPrompt = "You are a technical documentation assistant. Provide a very short, concise, one-paragraph overview (MAXIMUM 50 WORDS) describing the purpose, main responsibilities, and design of this C++ component. Do not include markdown code block styling or titles in your response; return ONLY the description paragraph. If the current description provided below is already good and accurate, please return it exactly as is.";
Juliet::String cotEnv = GetEnvironmentVar(arena, "ROMEO_ENABLE_COT");
bool enableCoT = (cotEnv.Size > 0 && (cotEnv.Data[0] == '1' || cotEnv.Data[0] == 't' || cotEnv.Data[0] == 'T' || cotEnv.Data[0] == 'y' || cotEnv.Data[0] == 'Y'));
const char* sysPromptNoCoT = "You are a technical documentation assistant. Provide a single, concise, unified paragraph overview (50-80 words) describing the main purpose and design of this C++ component.\n"
"Rules:\n"
"1. Do NOT split your output by header file or source file.\n"
"2. Do NOT include titles, markdown headers (e.g. # or ##), bullet lists, or code blocks.\n"
"3. Do NOT include <think> tags or reasoning text.\n"
"4. Return ONLY the description paragraph.";
const char* sysPromptCoT = "You are a technical documentation assistant. Provide a single, concise, unified paragraph overview (50-80 words) describing the main purpose and design of this C++ component.\n"
"Rules:\n"
"1. You may include your reasoning inside <think>...</think> tags.\n"
"2. Following the reasoning, output ONLY the final description paragraph.\n"
"3. Do NOT split your final output by header file or source file.";
const char* sysPrompt = enableCoT ? sysPromptCoT : sysPromptNoCoT;
size_t promptCap = strlen(sysPrompt) + entry.HeaderPath.Size + headerContent.Size + entry.CppPath.Size + cppContent.Size + entry.Description.Size + 512;
char* promptBuf = Juliet::ArenaPushArray<char>(arena, promptCap JULIET_DEBUG_PARAM("PromptBuf"));
@@ -229,9 +309,6 @@ namespace Romeo
Juliet::String prompt = { promptBuf, static_cast<size_t>(written > 0 ? written : 0) };
Juliet::String escapedPrompt = Json_EscapeString(arena, prompt);
// Try local Ollama
// Default Host: localhost:11434
// Default Model: qwen2.5-coder:1.5b
Juliet::String host = ConstString("127.0.0.1");
int port = 11434;
Juliet::String modelName = GetEnvironmentVar(arena, "ROMEO_LLM_MODEL");
@@ -240,34 +317,35 @@ namespace Romeo
modelName = ConstString("qwen2.5-coder:1.5b");
}
// Check / pull model if needed
if (!PullOllamaModelIfNeeded(arena, host, port, modelName))
{
return false;
}
printf("Romeo AI: Querying local Ollama (%.*s) for '%.*s' description...\n",
printf("Romeo AI: Querying local Ollama (%.*s) [CoT: %s] for '%.*s'...\n",
static_cast<int>(modelName.Size), CStr(modelName),
enableCoT ? "ON" : "OFF",
static_cast<int>(entry.HeaderPath.Size > 0 ? entry.HeaderPath.Size : entry.CppPath.Size),
CStr(entry.HeaderPath.Size > 0 ? entry.HeaderPath : entry.CppPath));
size_t bodyCap = modelName.Size + escapedPrompt.Size + 256;
int maxPredictTokens = enableCoT ? 600 : 300;
size_t bodyCap = modelName.Size + escapedPrompt.Size + 512;
char* bodyBuf = Juliet::ArenaPushArray<char>(arena, bodyCap JULIET_DEBUG_PARAM("OllamaBody"));
int bodyLen = snprintf(bodyBuf, bodyCap,
"{\"model\":\"%.*s\",\"prompt\":\"%.*s\",\"stream\":false}",
"{\"model\":\"%.*s\",\"prompt\":\"%.*s\",\"stream\":false,\"options\":{\"temperature\":0.2,\"num_predict\":%d}}",
static_cast<int>(modelName.Size), CStr(modelName),
static_cast<int>(escapedPrompt.Size), CStr(escapedPrompt)
static_cast<int>(escapedPrompt.Size), CStr(escapedPrompt),
maxPredictTokens
);
Juliet::String body = { bodyBuf, static_cast<size_t>(bodyLen > 0 ? bodyLen : 0) };
Juliet::String response = {};
bool success = PostWithRetry(
arena,
host,
port,
ConstString("/api/generate"),
false, // isHttps
false,
body,
response
);
@@ -280,7 +358,7 @@ namespace Romeo
Juliet::String desc = {};
if (Json_Query(root, "response", desc))
{
entry.Description = Juliet::StringCopy(arena, desc);
entry.Description = CleanAndFormatDescription(arena, desc, enableCoT);
return true;
}
}