Enrich Command (#75)

A command that is able to produce LLM embeddings in a couchbase lite database.

Co-authored-by: Alexander Baker <[email protected]>

Tool.cc

@@ -20,6 +20,7 @@
 #include "Logging.hh"
 #include "linenoise.h"
 #include "utf8.h"
+#include <charconv>
 #include <cstdio>
 #include <fstream>
 #include <regex>
@@ -266,3 +267,26 @@ alloc_slice Tool::readFile(const string &path) {
     in.read((char*)data.buf, size);
     return data;
 }
+
+int Tool::nextIntArg(const char *what, int minVal, int maxVal) {
+    return parseInt(nextArg(what), minVal, maxVal);
+}
+
+
+int Tool::parseInt(string_view str, int minVal, int maxVal) {
+    int value;
+    const char* end = str.data() + str.size();
+    auto [ptr, ec] = std::from_chars(str.data(), end, value);
+    const char* err = nullptr;
+    if (ec == errc::result_out_of_range)
+        err = " is out of range";
+    else if (ec != errc{} || ptr != end)
+        err = " is not a valid integer";
+    else if (value < minVal)
+        err = " is too small";
+    else if (value > maxVal)
+        err = " is too large";
+    if (err)
+        fail(string(str) + err);
+    return value;
+}

Tool.hh

@@ -25,6 +25,7 @@
 #include <string>
 #include <deque>
 #include <algorithm>
+#include <climits>
 
 #ifdef CMAKE
 #include "config.h"
@@ -176,6 +177,8 @@ public:
     std::string it(const char *str)          {return ansiItalic() + str + ansiReset();}
 
     std::string spaces(int n)                {return std::string(std::max(n, 1), ' ');}
+
+    int parseInt(std::string_view, int minVal = INT_MIN, int maxVal = INT_MAX);
 
 protected:
 
@@ -206,6 +209,8 @@ protected:
         return arg;
     }
 
+    int nextIntArg(const char *what, int minVal = INT_MIN, int maxVal = INT_MAX);
+
     /** If the next arg matches the given string, consumes it and returns true. */
     bool matchArg(const char *matchArg) {
         if (_argTokenizer.argument() != matchArg)
@@ -255,7 +260,15 @@ protected:
 
             if (flag == "--")
                 return;  // marks end of flags
-            if (!processFlag(flag, specs)) {
+
+            bool handled;
+            try {
+                handled = processFlag(flag, specs);
+            } catch (std::exception const& x) {
+                fail("in flag " + flag + ": " + x.what());
+            }
+
+            if (!handled) {
                 // Flags all subcommands accept:
                 if (flag == "--help") {
                     usage();

Xcode/Tools.xcodeproj/project.pbxproj

@@ -7,6 +7,7 @@
 	objects = {
 
 /* Begin PBXBuildFile section */
+		1A02703F2C2645A60025F2B5 /* EnrichCommand.cc in Sources */ = {isa = PBXBuildFile; fileRef = 1A02703D2C2645A60025F2B5 /* EnrichCommand.cc */; };
 		2716F94D2491822700BE21D9 /* CheckCommand.cc in Sources */ = {isa = PBXBuildFile; fileRef = 2716F94C2491822700BE21D9 /* CheckCommand.cc */; };
 		2716F95B2491857E00BE21D9 /* ReindexCommand.cc in Sources */ = {isa = PBXBuildFile; fileRef = 2716F95A2491857E00BE21D9 /* ReindexCommand.cc */; };
 		27175C0F261CE5F40045F3AC /* MkCollCommand.cc in Sources */ = {isa = PBXBuildFile; fileRef = 27175C0E261CE5F40045F3AC /* MkCollCommand.cc */; };
@@ -30,7 +31,6 @@
 		2751C83225D3650A00A9B39B /* linenoise.c in Sources */ = {isa = PBXBuildFile; fileRef = 2751C81C25D3650A00A9B39B /* linenoise.c */; };
 		2751C83325D3650A00A9B39B /* utf8.h in Headers */ = {isa = PBXBuildFile; fileRef = 2751C81F25D3650A00A9B39B /* utf8.h */; };
 		276CE5C9225FAA1600B681AC /* TokenizerTest.cc in Sources */ = {isa = PBXBuildFile; fileRef = 276CE5C8225FAA1600B681AC /* TokenizerTest.cc */; };
-		276CE5CC225FAC8500B681AC /* LibC++Debug.cc in Sources */ = {isa = PBXBuildFile; fileRef = 276CE5CB225FAC8500B681AC /* LibC++Debug.cc */; };
 		276CE5CF225FACDD00B681AC /* ArgumentTokenizer.cc in Sources */ = {isa = PBXBuildFile; fileRef = 27FC8DF722137C580083B033 /* ArgumentTokenizer.cc */; };
 		276CE5D1225FADB200B681AC /* tests_main.cc in Sources */ = {isa = PBXBuildFile; fileRef = 276CE5D0225FADB200B681AC /* tests_main.cc */; };
 		276D4AC527864A5500F61A89 /* MkIndexCommand.cc in Sources */ = {isa = PBXBuildFile; fileRef = 276D4AC427864A5500F61A89 /* MkIndexCommand.cc */; };
@@ -58,7 +58,6 @@
 		27FC8E64221381C60083B033 /* liblinenoise.a in Frameworks */ = {isa = PBXBuildFile; fileRef = 27FC8E5A221381890083B033 /* liblinenoise.a */; };
 		27FC8E67221383880083B033 /* libz.tbd in Frameworks */ = {isa = PBXBuildFile; fileRef = 27FC8E66221383880083B033 /* libz.tbd */; };
 		27FC8E6A221383AE0083B033 /* libLiteCoreREST-static.a in Frameworks */ = {isa = PBXBuildFile; fileRef = 27FC8E1A22137CB60083B033 /* libLiteCoreREST-static.a */; };
-		42030A6E24AC14F900283CE8 /* LibC++Debug.cc in Sources */ = {isa = PBXBuildFile; fileRef = 276CE5CB225FAC8500B681AC /* LibC++Debug.cc */; };
 		42030A7024AC152000283CE8 /* StringUtil.cc in Sources */ = {isa = PBXBuildFile; fileRef = 42030A6F24AC152000283CE8 /* StringUtil.cc */; };
 /* End PBXBuildFile section */
 
@@ -208,6 +207,7 @@
 /* End PBXCopyFilesBuildPhase section */
 
 /* Begin PBXFileReference section */
+		1A02703D2C2645A60025F2B5 /* EnrichCommand.cc */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.cpp.cpp; path = EnrichCommand.cc; sourceTree = "<group>"; };
 		2716F94C2491822700BE21D9 /* CheckCommand.cc */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.cpp.cpp; path = CheckCommand.cc; sourceTree = "<group>"; };
 		2716F95A2491857E00BE21D9 /* ReindexCommand.cc */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.cpp.cpp; path = ReindexCommand.cc; sourceTree = "<group>"; };
 		2716F9942493E59B00BE21D9 /* BUILDING.md */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = net.daringfireball.markdown; name = BUILDING.md; path = ../BUILDING.md; sourceTree = "<group>"; };
@@ -237,7 +237,6 @@
 		276CE5C0225FA9F400B681AC /* Tests */ = {isa = PBXFileReference; explicitFileType = "compiled.mach-o.executable"; includeInIndex = 0; path = Tests; sourceTree = BUILT_PRODUCTS_DIR; };
 		276CE5C8225FAA1600B681AC /* TokenizerTest.cc */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = TokenizerTest.cc; sourceTree = "<group>"; };
 		276CE5CA225FAA4C00B681AC /* tests.xcconfig */ = {isa = PBXFileReference; lastKnownFileType = text.xcconfig; path = tests.xcconfig; sourceTree = "<group>"; };
-		276CE5CB225FAC8500B681AC /* LibC++Debug.cc */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.cpp.cpp; path = "LibC++Debug.cc"; sourceTree = "<group>"; };
 		276CE5D0225FADB200B681AC /* tests_main.cc */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.cpp.cpp; path = tests_main.cc; sourceTree = "<group>"; };
 		276D4AC427864A5500F61A89 /* MkIndexCommand.cc */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.cpp.cpp; path = MkIndexCommand.cc; sourceTree = "<group>"; };
 		276D4AD22786502600F61A89 /* RmIndexCommand.cc */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.cpp.cpp; path = RmIndexCommand.cc; sourceTree = "<group>"; };
@@ -361,7 +360,6 @@
 			isa = PBXGroup;
 			children = (
 				276CE5C8225FAA1600B681AC /* TokenizerTest.cc */,
-				276CE5CB225FAC8500B681AC /* LibC++Debug.cc */,
 				276CE5D0225FADB200B681AC /* tests_main.cc */,
 			);
 			name = tests;
@@ -422,6 +420,7 @@
 				27FC8DD322137C330083B033 /* CpCommand.cc */,
 				273B20D3264B2D6900A14EC4 /* EditCommand.cc */,
 				271BA4AD227CC54300D49D13 /* EncryptCommand.cc */,
+				1A02703D2C2645A60025F2B5 /* EnrichCommand.cc */,
 				27FC8DD622137C330083B033 /* InfoCommand.cc */,
 				27FC8DD222137C330083B033 /* ListCommand.cc */,
 				27E95BAC24083D7C0013711C /* ListCommand.hh */,
@@ -760,7 +759,6 @@
 			files = (
 				272F00CF226F7FE500E62F72 /* cbl-log.cc in Sources */,
 				272F00D7226F956C00E62F72 /* LogDecoder_stub.cpp in Sources */,
-				42030A6E24AC14F900283CE8 /* LibC++Debug.cc in Sources */,
 				272F00D0226F7FFC00E62F72 /* Tool.cc in Sources */,
 				42030A7024AC152000283CE8 /* StringUtil.cc in Sources */,
 				272F00D1226F800000E62F72 /* ArgumentTokenizer.cc in Sources */,
@@ -772,7 +770,6 @@
 			buildActionMask = 2147483647;
 			files = (
 				276CE5D1225FADB200B681AC /* tests_main.cc in Sources */,
-				276CE5CC225FAC8500B681AC /* LibC++Debug.cc in Sources */,
 				276CE5CF225FACDD00B681AC /* ArgumentTokenizer.cc in Sources */,
 				276CE5C9225FAA1600B681AC /* TokenizerTest.cc in Sources */,
 			);
@@ -800,6 +797,7 @@
 				27FC8DDD22137C330083B033 /* ListCommand.cc in Sources */,
 				27FC8DF222137C490083B033 /* Endpoint.cc in Sources */,
 				27FC8DE522137C330083B033 /* SQLCommand.cc in Sources */,
+				1A02703F2C2645A60025F2B5 /* EnrichCommand.cc in Sources */,
 				27175C27261D097A0045F3AC /* MvCommand.cc in Sources */,
 				27E95BAB2408376B0013711C /* CBLiteCommand.cc in Sources */,
 				27175C21261D00200045F3AC /* CdCommand.cc in Sources */,
@@ -855,6 +853,19 @@
 			isa = XCBuildConfiguration;
 			baseConfigurationReference = 27FC8E4122137DAF0083B033 /* cblite.xcconfig */;
 			buildSettings = {
+				HEADER_SEARCH_PATHS = (
+					"$(inherited)",
+					"$(FLEECE)/API",
+					"$(FLEECE)/Fleece/Support",
+					"$(LITECORE)/C/include",
+					"$(LITECORE)/C/Cpp_include",
+					"$(LITECORE)/C",
+					"$(LITECORE)/Networking",
+					"$(LITECORE)/Replicator",
+					"$(LITECORE)/LiteCore/Support",
+					"$(LITECORE)/REST",
+					"$(LITECORE)/Networking/HTTP",
+				);
 			};
 			name = Debug_EE;
 		};
@@ -947,6 +958,19 @@
 			isa = XCBuildConfiguration;
 			baseConfigurationReference = 27FC8E4122137DAF0083B033 /* cblite.xcconfig */;
 			buildSettings = {
+				HEADER_SEARCH_PATHS = (
+					"$(inherited)",
+					"$(FLEECE)/API",
+					"$(FLEECE)/Fleece/Support",
+					"$(LITECORE)/C/include",
+					"$(LITECORE)/C/Cpp_include",
+					"$(LITECORE)/C",
+					"$(LITECORE)/Networking",
+					"$(LITECORE)/Replicator",
+					"$(LITECORE)/LiteCore/Support",
+					"$(LITECORE)/REST",
+					"$(LITECORE)/Networking/HTTP",
+				);
 			};
 			name = Release_EE;
 		};
@@ -1174,13 +1198,39 @@
 			isa = XCBuildConfiguration;
 			baseConfigurationReference = 27FC8E4122137DAF0083B033 /* cblite.xcconfig */;
 			buildSettings = {
+				HEADER_SEARCH_PATHS = (
+					"$(inherited)",
+					"$(FLEECE)/API",
+					"$(FLEECE)/Fleece/Support",
+					"$(LITECORE)/C/include",
+					"$(LITECORE)/C/Cpp_include",
+					"$(LITECORE)/C",
+					"$(LITECORE)/Networking",
+					"$(LITECORE)/Replicator",
+					"$(LITECORE)/LiteCore/Support",
+					"$(LITECORE)/REST",
+					"$(LITECORE)/Networking/HTTP",
+				);
 			};
 			name = Debug;
 		};
 		27FC8DD022137BF60083B033 /* Release */ = {
 			isa = XCBuildConfiguration;
 			baseConfigurationReference = 27FC8E4122137DAF0083B033 /* cblite.xcconfig */;
 			buildSettings = {
+				HEADER_SEARCH_PATHS = (
+					"$(inherited)",
+					"$(FLEECE)/API",
+					"$(FLEECE)/Fleece/Support",
+					"$(LITECORE)/C/include",
+					"$(LITECORE)/C/Cpp_include",
+					"$(LITECORE)/C",
+					"$(LITECORE)/Networking",
+					"$(LITECORE)/Replicator",
+					"$(LITECORE)/LiteCore/Support",
+					"$(LITECORE)/REST",
+					"$(LITECORE)/Networking/HTTP",
+				);
 			};
 			name = Release;
 		};

cblite/CBLiteCommand.hh

@@ -146,6 +146,7 @@ CBLiteCommand* newCheckCommand(CBLiteTool&);
 CBLiteCommand* newCompactCommand(CBLiteTool&);
 CBLiteCommand* newCpCommand(CBLiteTool&);
 CBLiteCommand* newEditCommand(CBLiteTool&);
+CBLiteCommand* newEnrichCommand(CBLiteTool&);
 CBLiteCommand* newExportCommand(CBLiteTool&);
 CBLiteCommand* newImportCommand(CBLiteTool&);
 CBLiteCommand* newInfoCommand(CBLiteTool&);

cblite/CBLiteTool.cc

@@ -208,6 +208,10 @@ void CBLiteTool::openDatabase(string pathStr, bool interactive) {
     C4Error err;
     const C4Error kEncryptedDBError = {LiteCoreDomain, kC4ErrorNotADatabaseFile};
 
+    if (const char* extPath = getenv("CBLITE_EXTENSION_PATH")) {
+        c4_setExtensionPath(slice(extPath));
+    }
+
     if (!_dbNeedsPassword) {
         _db = c4db_openNamed(slice(dbName), &config, &err);
     } else {
@@ -287,6 +291,7 @@ static constexpr struct {const char* name; ToolFactory factory;} kSubcommands[]
     {"compact", newCompactCommand},
     {"cp",      newCpCommand},
     {"edit",    newEditCommand},
+    {"enrich",  newEnrichCommand},
     {"export",  newExportCommand},
     {"file",    newInfoCommand},
     {"get",     newCatCommand},

cblite/CMakeLists.txt

@@ -52,6 +52,7 @@ target_include_directories(
     ${LITECORE}Networking/
     ${LITECORE}Networking/HTTP
     ${LITECORE}Replicator
+    ${LITECORE}REST/
     ${LITECORE}vendor/fleece/API/
     ${LITECORE}vendor/fleece/Fleece/Support/ # PlatformCompat.hh
     ${CMAKE_BINARY_DIR}/generated_headers/

cblite/CompactCommand.cc

@@ -44,7 +44,7 @@ class CompactCommand : public CBLiteCommand {
     void runSubcommand() override {
         // Read params:
         processFlags({
-            {"--prune",         [&]{_prune = stoi(nextArg("depth for --prune"));}},
+            {"--prune",         [&]{_prune = nextIntArg("depth for --prune", 1);}},
             {"--purgeDeleted",  [&]{_purgeDeleted = true;}},
         });
 
@@ -177,14 +177,14 @@ class CompactCommand : public CBLiteCommand {
                 if (c4doc_selectCommonAncestorRevision(doc, doc->selectedRev.revID, currentRevID))
                     branchPoint = doc->selectedRev.revID;
                 // First count the number of revs on the branch:
-                c4doc_selectRevision(doc, closedBranch, false, nullptr);
+                [[maybe_unused]] bool _ = c4doc_selectRevision(doc, closedBranch, false, nullptr);
                 do {
                     ++nPrunedRevs;
                     if (doc->selectedRev.flags & kRevKeepBody)
                         ++nRemovedBodies;
                 } while (c4doc_selectParentRevision(doc) && doc->selectedRev.revID != branchPoint);
                 // Then prune the entire branch:
-                c4doc_purgeRevision(doc, closedBranch, nullptr);
+                _ = c4doc_purgeRevision(doc, closedBranch, nullptr);
             } else {
                 // Walk its ancestor chain, counting how many revs are deeper than maxDepth:
                 unsigned branchDepth = 1, keepBodyDepth = 0;

cblite/DocBranchIterator.hh

@@ -24,7 +24,7 @@ public:
     }
 
     DocBranchIterator& operator++() {
-        c4doc_selectRevision(_doc, _branchID, false, nullptr);
+        [[maybe_unused]] bool _ = c4doc_selectRevision(_doc, _branchID, false, nullptr);
         _branchID = fleece::nullslice;
         while (c4doc_selectNextRevision(_doc)) {
             if (_doc->selectedRev.flags & kRevLeaf) {