Reentrant library API

.github/workflows/ci.yml

@@ -0,0 +1,158 @@
+name: CI
+
+on:
+  push:
+    branches: [ library-api, GoogleImport, master ]
+  pull_request:
+    branches: [ library-api, GoogleImport, master ]
+
+jobs:
+
+  # ── Linux ──────────────────────────────────────────────────────────────────
+  linux:
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        cc: [gcc, clang]
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Install dependencies
+        run: sudo apt-get update && sudo apt-get install -y zlib1g-dev
+
+      - name: Build CLI + library
+        run: make CC=${{ matrix.cc }}
+
+      - name: Build static library
+        run: make libprodigal.a CC=${{ matrix.cc }}
+
+      - name: Build and run tests
+        run: make test CC=${{ matrix.cc }}
+
+      - name: Regression - metagenomic GFF
+        run: |
+          ./prodigal -i anthus_aco.fas -p meta -f gff -q -o /tmp/meta.gff
+          diff testdata/ground_truth/ref_meta.gff /tmp/meta.gff
+
+      - name: Regression - metagenomic GBK
+        run: |
+          ./prodigal -i anthus_aco.fas -p meta -f gbk -q -o /tmp/meta.gbk
+          diff testdata/ground_truth/ref_meta.gbk /tmp/meta.gbk
+
+      - name: Regression - metagenomic SCO
+        run: |
+          ./prodigal -i anthus_aco.fas -p meta -f sco -q -o /tmp/meta.sco
+          diff testdata/ground_truth/ref_meta.sco /tmp/meta.sco
+
+      - name: Regression - metagenomic proteins
+        run: |
+          ./prodigal -i anthus_aco.fas -p meta -f gff -q -o /dev/null \
+            -a /tmp/meta.proteins
+          diff testdata/ground_truth/ref_meta.proteins /tmp/meta.proteins
+
+      - name: Regression - metagenomic nucleotides
+        run: |
+          ./prodigal -i anthus_aco.fas -p meta -f gff -q -o /dev/null \
+            -d /tmp/meta.nucl
+          diff testdata/ground_truth/ref_meta.nucl /tmp/meta.nucl
+
+      - name: Regression - single genome GFF
+        run: |
+          ./prodigal -i anthus_aco.fas -p single -f gff -q -o /tmp/single.gff
+          diff testdata/ground_truth/ref_single.gff /tmp/single.gff
+
+      - name: Regression - training file round-trip
+        run: |
+          ./prodigal -i anthus_aco.fas -t /tmp/train.bin -q
+          diff testdata/ground_truth/ref_train.bin /tmp/train.bin
+          ./prodigal -i anthus_aco.fas -t /tmp/train.bin -f gff -q \
+            -o /tmp/trained.gff
+          diff testdata/ground_truth/ref_trained.gff /tmp/trained.gff
+
+  # ── macOS ──────────────────────────────────────────────────────────────────
+  # Reference binaries were generated on Linux x86_64.  struct _training has
+  # different padding on ARM64, so training-file comparisons use self-
+  # consistent round-trips rather than cross-platform reference files.
+  # Text output (GFF, GBK, etc.) is architecture-independent.
+  macos:
+    runs-on: macos-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Build CLI + library
+        run: make
+
+      - name: Build and run tests
+        run: make test
+
+      - name: Regression - metagenomic GFF
+        run: |
+          ./prodigal -i anthus_aco.fas -p meta -f gff -q -o /tmp/meta.gff
+          diff testdata/ground_truth/ref_meta.gff /tmp/meta.gff
+
+      - name: Regression - single genome GFF
+        run: |
+          ./prodigal -i anthus_aco.fas -p single -f gff -q -o /tmp/single.gff
+          diff testdata/ground_truth/ref_single.gff /tmp/single.gff
+
+      - name: Regression - training file round-trip (self-consistent)
+        run: |
+          ./prodigal -i anthus_aco.fas -t /tmp/train1.bin -q
+          ./prodigal -i anthus_aco.fas -t /tmp/train1.bin -f gff -q \
+            -o /tmp/trained1.gff
+          ./prodigal -i anthus_aco.fas -t /tmp/train2.bin -q
+          diff /tmp/train1.bin /tmp/train2.bin
+          ./prodigal -i anthus_aco.fas -t /tmp/train2.bin -f gff -q \
+            -o /tmp/trained2.gff
+          diff /tmp/trained1.gff /tmp/trained2.gff
+
+  # ── WASM (library only, via Emscripten) ────────────────────────────────────
+  wasm:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Setup Emscripten
+        uses: mymindstorm/setup-emsdk@v14
+
+      - name: Build library with emcc
+        run: |
+          CORE="bitmap.c dprog.c gene.c metagenomic.c node.c sequence.c training.c prodigal_api.c"
+          emcc -O3 -Wall -DPRODIGAL_NO_MAIN -c $CORE
+          emar rcs libprodigal.a *.o
+          echo "WASM library build succeeded"
+
+      - name: Build and run test suite in Node.js
+        run: |
+          CORE="bitmap.c dprog.c gene.c metagenomic.c node.c sequence.c training.c prodigal_api.c"
+          emcc -O3 -Wall -DPRODIGAL_NO_MAIN $CORE test_api.c -o test_api.js \
+            -s ALLOW_MEMORY_GROWTH=1 \
+            -s INITIAL_MEMORY=134217728 \
+            -lm
+          node test_api.js
+
+  # ── Windows (MSYS2/MinGW-w64) ─────────────────────────────────────────────
+  windows:
+    runs-on: windows-latest
+    defaults:
+      run:
+        shell: msys2 {0}
+    steps:
+      - uses: actions/checkout@v4
+
+      - uses: msys2/setup-msys2@v2
+        with:
+          msystem: UCRT64
+          install: mingw-w64-ucrt-x86_64-gcc make mingw-w64-ucrt-x86_64-zlib diffutils
+
+      - name: Build CLI + library
+        run: make
+
+      - name: Build and run tests
+        run: make test
+
+      - name: Regression - metagenomic GFF (determinism)
+        run: |
+          ./prodigal -i anthus_aco.fas -p meta -f gff -q -o /tmp/meta1.gff
+          ./prodigal -i anthus_aco.fas -p meta -f gff -q -o /tmp/meta2.gff
+          diff /tmp/meta1.gff /tmp/meta2.gff

.gitignore

@@ -1,2 +1,7 @@
 *.o
+*.pic.o
 prodigal
+prodigal_native
+libprodigal.a
+libprodigal.so
+test_api

CHANGES

@@ -1,3 +1,8 @@
+VERSION 2.6.4
+
+* Added a reentrant embeddable API
+* Added basic CI testing (Linux, macOS, Windows, WASM)
+
 VERSION 2.6.3
 
 * Fixed a bug in protein translation output of partial genes where TTG/GTG

Makefile

@@ -25,33 +25,73 @@ CFLAGS  += -pedantic -Wall -O3 -DSUPPORT_GZIP_COMPRESSED
 LFLAGS = -lm $(LDFLAGS) -lz
 
 TARGET  = prodigal
-ZTARGET  = zprodigal
-SOURCES = $(shell echo *.c)
+INSTALLDIR  = /usr/local/bin
+
+# Source file groups
+CORE_SOURCES = bitmap.c dprog.c gene.c metagenomic.c node.c sequence.c training.c
+API_SOURCES  = prodigal_api.c
+CLI_SOURCE   = main.c
+TEST_SOURCE  = test_api.c
+
 HEADERS = $(shell echo *.h)
-OBJECTS = $(SOURCES:.c=.o)
-ZOBJECTS = $(SOURCES:.c=.oz)
 
-INSTALLDIR  = /usr/local/bin
+# CLI objects (no PRODIGAL_NO_MAIN)
+CLI_OBJ = $(CLI_SOURCE:.c=.o)
+CLI_CORE_OBJS = $(CORE_SOURCES:.c=.o)
+CLI_API_OBJ = $(API_SOURCES:.c=.o)
 
-all: $(TARGET)
+# Library objects (with PRODIGAL_NO_MAIN to suppress stderr writes)
+LIB_CORE_OBJS = $(CORE_SOURCES:.c=.lib.o)
+LIB_API_OBJ = $(API_SOURCES:.c=.lib.o)
+LIB_OBJS = $(LIB_CORE_OBJS) $(LIB_API_OBJ)
 
-$(TARGET): $(OBJECTS)
-	$(CC) $(CFLAGS) -o $@ $^ $(LFLAGS)
+# Default: build CLI binary
+all: $(TARGET)
 
+# CLI objects: compiled WITHOUT PRODIGAL_NO_MAIN
 %.o: %.c $(HEADERS)
 	$(CC) $(CFLAGS) -c -o $@ $<
 
+# Library objects: compiled WITH PRODIGAL_NO_MAIN (suppresses stderr in core)
+%.lib.o: %.c $(HEADERS)
+	$(CC) $(CFLAGS) -DPRODIGAL_NO_MAIN -c -o $@ $<
+
+# Static library
+libprodigal.a: $(LIB_OBJS)
+	ar rcs $@ $^
+
+# CLI binary: link CLI main.o + core objects (not using lib to keep stderr)
+$(TARGET): $(CLI_OBJ) $(CLI_CORE_OBJS) $(CLI_API_OBJ)
+	$(CC) $(CFLAGS) -o $@ $^ $(LFLAGS)
+
+# PIC objects for shared library
+%.pic.o: %.c $(HEADERS)
+	$(CC) $(CFLAGS) -fPIC -DPRODIGAL_NO_MAIN -c -o $@ $<
+
+LIB_PIC_OBJS = $(CORE_SOURCES:.c=.pic.o) $(API_SOURCES:.c=.pic.o)
+
+# Shared library
+libprodigal.so: $(LIB_PIC_OBJS)
+	$(CC) -shared -o $@ $^ -lm
+
+# Test runner (linked against static library)
+test_api: $(TEST_SOURCE) libprodigal.a
+	$(CC) $(CFLAGS) -DPRODIGAL_NO_MAIN -o $@ $< -L. -lprodigal $(LFLAGS)
+
+test: test_api
+	./test_api
+
 install: $(TARGET)
 	install -d -m 0755 $(INSTALLDIR)
 	install -m 0755 $(TARGET) $(INSTALLDIR)
- 
+
 uninstall:
 	-rm $(INSTALLDIR)/$(TARGET)
 
 clean:
-	-rm -f $(OBJECTS) $(ZOBJECTS)
- 
+	-rm -f *.o *.lib.o *.pic.o
+
 distclean: clean
-	-rm -f $(TARGET)
+	-rm -f $(TARGET) libprodigal.a libprodigal.so test_api
 
-.PHONY: all install uninstall clean distclean
+.PHONY: all install uninstall clean distclean test

README.md

@@ -8,6 +8,9 @@ prodigal -i my.metagenome.fna -o my.genes -a my.proteins.faa -p meta
 prodigal -h
 ```
 
+### New in 2.6.4 (April 2026)
+  * Reentrant API created
+
 ### New in 2.6.3 (February 2016)
   * Fixed a bug in protein translation output of partial genes where TTG/GTG
 codons were being incorrectly translated to methionine.

VERSION

@@ -1 +1 @@
-v2.6.3: February 2016
+v2.6.4: April 2026

gene.c

@@ -51,7 +51,9 @@ int add_genes(struct _gene *glist, struct _node *nod, int dbeg) {
     }
     path = nod[path].tracef;
     if(ctr == MAX_GENES) {
+#ifndef PRODIGAL_NO_MAIN
       fprintf(stderr, "warning, max # of genes exceeded, truncating...\n");
+#endif
       return ctr;
     }
   }