make_scripts toegvoegd

bookscraper/scraper/book_scraper.py

@@ -85,6 +85,8 @@ class BookScraper:
         else:
             self.download_all()
 
+        self.prepare_scripts(self.base_path)
+
         return {"title": self.book_title}
 
     # ------------------------------------------------------------
@@ -345,7 +347,17 @@ class BookScraper:
             time.sleep(wait)
             attempt += 1
 
-        fname.write_text(ch.text, encoding="utf-8")
+        if ch.number == 1:
+            header = (
+                f"Description:\n{self.book_description}\n"
+                f" {ch.url}\n"
+                "----------------------------------------\n\n"
+            )
+            content = header + ch.text
+        else:
+            content = ch.text
+
+        fname.write_text(content, encoding="utf-8")
         log_debug(f"Saved chapter to v{volume}: {fname}")
         chapter_delay = float(os.getenv("CHAPTER_DELAY", "2"))
         log_debug(f"Throttling {chapter_delay}s before next chapter")
@@ -389,3 +401,126 @@ class BookScraper:
         raw = "\n".join(parts)
         raw = clean_text(raw, self.site.replacements)
         return raw.strip()
+    # ------------------------------------------------------------
+    # SCRIPT GENERATOR (C# prepareScripts equivalent)
+    # ------------------------------------------------------------
+
+    def prepare_scripts(self, path: Path):
+        log_debug("Preparing scripts...")
+
+        # Alleen volume folders (v1, v2, v3, ...)
+        dirs = sorted([
+            d for d in path.iterdir()
+            if d.is_dir() and d.name.startswith("v")
+        ])
+
+        # --------------------------------------------------------
+        # M4B MERGE COMMANDS
+        # --------------------------------------------------------
+        m4b_commands = []
+        for d in dirs:
+            volname = d.name
+            cmd = (
+                'm4b-tool merge --jobs=4 '
+                f'--writer="{self.book_author}" '
+                f'--albumartist="{self.book_author}" '
+                f'--album="{self.book_title}" '
+                f'--name="{self.book_title}" '
+                f'--output-file="{self.book_title}-{volname}.m4b" "./{volname}" -vvv'
+            )
+            m4b_commands.append(cmd)
+
+        # Mooie formatting
+        m4b_joined = " \\\n    && ".join(m4b_commands) + "\n\n"
+
+        move_script = r'''
+    find ./ -maxdepth 1 -name "*.m4b" | while read fname; do
+        echo "moving $(ls -lah "$fname")"
+        mv "$fname" ../
+    done
+    '''.strip() + "\n"
+
+        m4b_joined += move_script
+
+        # --------------------------------------------------------
+        # SAY TTS SCRIPT (jouw verbeterde versie – identiek voor say.txt en allinone.txt)
+        # --------------------------------------------------------
+        say_script = r'''#!/usr/bin/env bash
+
+    set -euo pipefail
+
+    main_dir="$( cd "$( dirname "$0" )" && pwd )"
+
+    shopt -s nocasematch    # case-insensitive matching
+
+    echo "=== TTS START ==="
+    echo "Main directory: $main_dir"
+    echo
+
+    for subfolder in "$main_dir"/*; do
+        if [[ -d "$subfolder" ]]; then
+
+            audiofolder="$subfolder/Audio"
+            mkdir -p "$audiofolder"
+
+            for entry in "$subfolder"/*.txt; do
+                [[ -f "$entry" ]] || continue
+
+                fn=$(basename "$entry")
+                inputfile="$entry"
+                outputfile="$audiofolder/${fn%.*}.m4b"
+
+                now=$(date +"%T")
+
+                echo "[$now] Processing $fn"
+                echo "Input : $inputfile"
+                echo "Output: $outputfile"
+
+                if [[ -f "$outputfile" ]]; then
+                    echo "[$now] EXISTS — skipping"
+                else
+                    say --voice=Sinji \
+                        --output-file="$outputfile" \
+                        --input-file="$inputfile" \
+                        --file-format=m4bf \
+                        --quality=127 \
+                        -r 200 \
+                        --data-format=aac
+                fi
+
+                echo
+            done
+        fi
+    done
+
+    echo "=== CLEANUP TINY FILES ==="
+    find "$main_dir" -name "*.m4b" -size -580c | while read -r fname; do
+        echo "Deleting: $(ls -lah "$fname")"
+        rm "$fname"
+    done
+    echo
+    '''
+
+        # --------------------------------------------------------
+        # SCHRIJF BESTANDEN WEG
+        # --------------------------------------------------------
+
+        # allinone.txt = TTS + merge
+        (path / "allinone.txt").write_text(
+            say_script + "\n" + m4b_joined,
+            encoding="utf-8"
+        )
+
+        # makebook.txt = alleen merge
+        (path / "makebook.txt").write_text(
+            m4b_joined,
+            encoding="utf-8"
+        )
+
+        # say.txt = alleen TTS (jouw vraag!)
+        (path / "say.txt").write_text(
+            say_script,
+            encoding="utf-8"
+        )
+
+        log_debug("Script files written: allinone.txt, makebook.txt, say.txt")