From 3263fcbb58566fc1fe1ec3891fccd858cb4e6894 Mon Sep 17 00:00:00 2001
From: Holden Rohrer <hr@hrhr.dev>
Date: Sun, 24 May 2020 01:36:32 -0400
Subject: added the framework for reading

---
 read.c | 114 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 read.h |   9 ++++++
 2 files changed, 123 insertions(+)
 create mode 100644 read.c
 create mode 100644 read.h

diff --git a/read.c b/read.c
new file mode 100644
index 0000000..e896863
--- /dev/null
+++ b/read.c
@@ -0,0 +1,114 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdbool.h>
+#include <string.h>
+
+#include "read.h"
+#include "strbst.h"
+#include "nodelink.h"
+#include "ll.h"
+
+void consumespaces(FILE* file) {
+    int c;
+    while ( ( c = fgetc(file) ) != EOF && c != ' ' );
+    ungetc(c, file);
+}
+
+void wsclean(char* str, size_t sz) { // 
+    bool nl = false;
+    if (str[sz] == '\n') {
+        sz--;
+        nl = true;
+    }
+    size_t i;
+    for (i = sz-1; i >= 0 && str[i] == ' '; i--);
+    if (nl) {
+        str[i+1] = '\n';
+        str[i+2] = 0;
+    } else {
+        str[i+1] = 0;
+    }
+
+}
+
+char* empty(void) {
+    char* str = malloc(sizeof(char));
+    *str = 0;
+    return str;
+}
+
+char* line(FILE* file) {
+    char* link = NULL;
+    size_t n;
+    getline(&link, &n, file);
+    wsclean(link, n);
+    if (*link == '\n') {
+        free(link);
+        link = empty();
+    }
+    return link;
+}
+
+link* insorget(strbst* tgt, char* name) {
+    link* get = query(tgt, name);
+    if (!(*get)) {
+        get = newlink(newnode());
+        insbst(tgt, name, get);
+    }
+    return get;
+}
+
+char* getdesc(FILE* file) {
+    int c;
+    size_t sz;
+    llnode *head, *tail;
+    head = tail = appendll(NULL, NULL);
+    while ( (c = getc()) != EOF && c != ':' && c != '-' ) {
+        tail = tail->next = appendll(tail, line());
+        sz += strlen(tail->str); // this should be traced but I haven't
+    }
+    if (c != EOF) ungetc(c, file);
+    tail = head->next; free(head);
+    char *out, *pos;
+    pos = out = malloc(sizeof(char)*(sz+1)); out[sz] = 0;
+    while (tail != NULL) {
+        size_t len = strlen(tail->str);
+        memcpy(pos, tail->str, len);
+        pos += len;
+        llnode* tmp = tail;
+        tail = tail->next;
+        free(tmp); free(tmp->str);
+    }
+    return pos;
+}
+
+node* readfile(char* name) {
+    FILE* read = fopen(name, "r");
+    node *root, *cur;
+    cur = root = newnode();
+    link* curl = insbst(root->links, name, root);
+    // getline
+    while ( ( int c = fgetc(read) ) != EOF ) {
+        switch (c) {
+            case ':':
+                consumespaces(file);
+                curl = insorget(root->links, line());
+                cur = curl->to;
+                break;
+
+            case '-':
+                consumespaces(file);
+                char* lname = line();
+                curl = insbst(cur->links, lname,
+                              insorget(root->links, lname)->to);
+                break;
+
+            default:
+                ungetc(c, read);
+                curl->desc = getdesc();
+                break;
+        }
+    }
+    fclose(read);
+    return root;
+}
diff --git a/read.h b/read.h
new file mode 100644
index 0000000..c01d4e6
--- /dev/null
+++ b/read.h
@@ -0,0 +1,9 @@
+#ifndef __READ_H__
+#define __READ_H__
+
+#include "strbst.h"
+#include "nodelink.h"
+
+node* readfile(char* name); // returns a root node with all as children
+
+#endif
-- 
cgit