basal-agent / src / sensor.h

/* sensor.h — filesystem perception: TWO cheap features (no open) + costly observation. */
#ifndef BASAL_SENSOR_H
#define BASAL_SENSOR_H

#include <stddef.h>

/* The thing predicted: is this file hand-written SOURCE or a build artifact
 * (GENERATED)? Determined by whether any ANCESTOR directory is a build marker
 * (node_modules, dist, __pycache__, ...) — context the two cheap features
 * (extension + immediate-parent name) cannot see directly. */
enum { CLASS_SOURCE = 0, CLASS_GENERATED = 1, NUM_CLASSES = 2 };

/* One file. Two features knowable WITHOUT opening: extension and parent-dir name.
 * `klass` is the observation, computed lazily on the first open. */
typedef struct {
	char  *path;
	int    f_ext;     /* extension feature id */
	int    f_dir;     /* parent-directory-name feature id */
	int    klass;     /* -1 until opened */
} File;

/* A growing table mapping strings <-> small integer feature ids. */
typedef struct {
	char **names;
	size_t count, cap;
} FeatureTable;

typedef struct {
	File         *files;
	size_t        count, cap;
	FeatureTable  ext;   /* extension vocabulary */
	FeatureTable  dir;   /* parent-dir-name vocabulary */
} Corpus;

int   corpus_collect(Corpus *c, const char *root);
void  corpus_free(Corpus *c);
int   sensor_open(Corpus *c, size_t i);          /* the costly act */

const char *ext_name(const Corpus *c, int f);
const char *dir_name(const Corpus *c, int f);

#endif