streamdown 0.17.0__tar.gz → 0.19.0__tar.gz

{streamdown-0.17.0 → streamdown-0.19.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: streamdown
-Version: 0.17.0
+Version: 0.19.0
 Summary: A streaming markdown renderer for modern terminals with syntax highlighting
 Project-URL: Homepage, https://github.com/kristopolous/Streamdown
 Project-URL: Bug Tracker, https://github.com/kristopolous/Streamdown/issues
@@ -31,23 +31,25 @@ Description-Content-Type: text/markdown
 <img src=https://github.com/user-attachments/assets/0468eac0-2a00-4e98-82ca-09e6ac679357/>
 <br/>
 <a href=https://pypi.org/project/streamdown><img src=https://badge.fury.io/py/streamdown.svg/></a>
+<br/><strong>Terminal streaming markdown that rocks</strong>
+
 </p>
 
-**The streaming markdown renderer for the terminal that rocks**
 
-Streamdown works with [simonw's llm](https://github.com/simonw/llm) along with any other streaming markdown. You even get full readline and keyboard navigation support.
+Streamdown works with [simonw's llm](https://github.com/simonw/llm) along with any other streaming markdown, even something basic like curl. 
+It supports standard piping like any normal pager and a clean `execvp` option for robustly wrapping around interactive programs with readline or their own ANSI stuff to manage.
 ```bash
 $ pip install streamdown
 ```
 ![Streamdown is Amazing](https://github.com/user-attachments/assets/268cb340-78cc-4df0-a773-c5ac95eceeeb)
 
 ### Provides clean copyable code for long code lines
-Some *inferior* renderers inject line breaks when copying code that wraps around. We're better and now you are too!
+Other renderers inject line breaks when copying code that wraps around. We're better and now you are too!
 ![Handle That Mandle](https://github.com/user-attachments/assets/a27aa70c-f691-4796-84f0-c2eb18c7de23)
 **Tip**: You can make things prettier if you don't mind if this guarantee is broken. See the `PrettyBroken` flag below!
 
 ### Supports images
-Here's kitty and alacritty. Try to do that in glow...
+Here's kitty and alacritty. 
 ![doggie](https://github.com/user-attachments/assets/81c43983-68cd-40c1-b1d5-aa3a52004504)
 
 ### Supports hyperlinks (OSC 8) and clipboard (OSC 52)
@@ -56,9 +58,13 @@ Here's kitty and alacritty. Try to do that in glow...
 ### Supports tables
 ![table](https://github.com/user-attachments/assets/dbe3d13e-6bac-4f45-bf30-f1857ed98898)
 
-As well as everything else...
+#### As well as everything else...
 ![dunder](https://github.com/user-attachments/assets/d41d7fec-6dec-4387-b53d-f2098f269a5e)
 
+#### ...even CJK 
+Compare how streamdown wraps around and spaces this tabular Chinese description of programming languages to the same file using glow.
+![cjk](https://github.com/user-attachments/assets/b831a5f3-7ef0-48b8-8d17-bb7b605df16a)
+
 ### Colors are highly (and quickly) configurable for people who care a lot, or just a little.
 ![configurable](https://github.com/user-attachments/assets/19ca2ec9-8ea1-4a79-87ca-8352789269fe)
 

{streamdown-0.17.0 → streamdown-0.19.0}/README.md

@@ -2,23 +2,25 @@
 <img src=https://github.com/user-attachments/assets/0468eac0-2a00-4e98-82ca-09e6ac679357/>
 <br/>
 <a href=https://pypi.org/project/streamdown><img src=https://badge.fury.io/py/streamdown.svg/></a>
+<br/><strong>Terminal streaming markdown that rocks</strong>
+
 </p>
 
-**The streaming markdown renderer for the terminal that rocks**
 
-Streamdown works with [simonw's llm](https://github.com/simonw/llm) along with any other streaming markdown. You even get full readline and keyboard navigation support.
+Streamdown works with [simonw's llm](https://github.com/simonw/llm) along with any other streaming markdown, even something basic like curl. 
+It supports standard piping like any normal pager and a clean `execvp` option for robustly wrapping around interactive programs with readline or their own ANSI stuff to manage.
 ```bash
 $ pip install streamdown
 ```
 ![Streamdown is Amazing](https://github.com/user-attachments/assets/268cb340-78cc-4df0-a773-c5ac95eceeeb)
 
 ### Provides clean copyable code for long code lines
-Some *inferior* renderers inject line breaks when copying code that wraps around. We're better and now you are too!
+Other renderers inject line breaks when copying code that wraps around. We're better and now you are too!
 ![Handle That Mandle](https://github.com/user-attachments/assets/a27aa70c-f691-4796-84f0-c2eb18c7de23)
 **Tip**: You can make things prettier if you don't mind if this guarantee is broken. See the `PrettyBroken` flag below!
 
 ### Supports images
-Here's kitty and alacritty. Try to do that in glow...
+Here's kitty and alacritty. 
 ![doggie](https://github.com/user-attachments/assets/81c43983-68cd-40c1-b1d5-aa3a52004504)
 
 ### Supports hyperlinks (OSC 8) and clipboard (OSC 52)
@@ -27,9 +29,13 @@ Here's kitty and alacritty. Try to do that in glow...
 ### Supports tables
 ![table](https://github.com/user-attachments/assets/dbe3d13e-6bac-4f45-bf30-f1857ed98898)
 
-As well as everything else...
+#### As well as everything else...
 ![dunder](https://github.com/user-attachments/assets/d41d7fec-6dec-4387-b53d-f2098f269a5e)
 
+#### ...even CJK 
+Compare how streamdown wraps around and spaces this tabular Chinese description of programming languages to the same file using glow.
+![cjk](https://github.com/user-attachments/assets/b831a5f3-7ef0-48b8-8d17-bb7b605df16a)
+
 ### Colors are highly (and quickly) configurable for people who care a lot, or just a little.
 ![configurable](https://github.com/user-attachments/assets/19ca2ec9-8ea1-4a79-87ca-8352789269fe)
 

{streamdown-0.17.0 → streamdown-0.19.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "streamdown"
-version = "0.17.0"
+version = "0.19.0"
 description = "A streaming markdown renderer for modern terminals with syntax highlighting"
 readme = "README.md"
 requires-python = ">=3.8"

streamdown-0.19.0/ss-new.py

@@ -0,0 +1,45 @@
+import os
+import pty
+import sys
+import tty
+import termios
+import select
+import re
+
+def main():
+    orig_attrs = termios.tcgetattr(sys.stdin)
+    try:
+        tty.setraw(sys.stdin)
+        pid, fd = pty.fork()
+
+        if pid == 0:
+            # Child process: run any command
+            os.execvp("llm", ["llm", "chat"])
+        else:
+            buffer = b""
+
+            while True:
+                r, _, _ = select.select([fd, sys.stdin], [], [])
+
+                if sys.stdin in r:
+                    data = os.read(sys.stdin.fileno(), 1024)
+                    if not data:
+                        break
+                    os.write(fd, data)
+
+                if fd in r:
+                    data = os.read(fd, 1024)
+                    if not data:
+                        break
+
+                    buffer += data
+                    # Replace "fizz" only in printable content
+                    output = re.sub(rb'fizz', b'fizzbuzz', buffer)
+                    os.write(sys.stdout.fileno(), output)
+                    buffer = b""
+    finally:
+        termios.tcsetattr(sys.stdin, termios.TCSADRAIN, orig_attrs)
+
+if __name__ == "__main__":
+    main()
+

streamdown-0.19.0/ss.py

@@ -0,0 +1,69 @@
+import os
+import pty
+import select
+import sys
+import tty
+import termios
+import fcntl
+import struct
+
+def set_pty_size(fd, target_fd):
+    """Set window size of PTY to match the real terminal."""
+    s = fcntl.ioctl(target_fd, termios.TIOCGWINSZ, b"\x00" * 8)
+    fcntl.ioctl(fd, termios.TIOCSWINSZ, s)
+
+def main():
+    # Save original terminal settings
+    orig_attrs = termios.tcgetattr(sys.stdin.fileno())
+    try:
+        tty.setraw(sys.stdin.fileno())  # raw mode to send Ctrl-C, etc.
+        pid, fd = pty.fork()
+
+        if pid == 0:
+            os.execvp("bash", ["bash"])
+        else:
+            set_pty_size(fd, sys.stdin.fileno())
+
+            while True:
+                r, _, _ = select.select([fd, sys.stdin], [], [])
+
+                if sys.stdin in r:
+                    user_input = os.read(sys.stdin.fileno(), 1024)
+                    if not user_input:
+                        break
+                    os.write(fd, user_input)
+
+                if fd in r:
+                    output = os.read(fd, 1024)
+                    if not output:
+                        break
+
+                    # Carefully handle ANSI sequences
+                    # Split on ANSI escape sequences to avoid breaking them
+                    chunks = []
+                    i = 0
+                    while i < len(output):
+                        if output[i:i+1] == b'\x1b':
+                            end = i + 1
+                            while end < len(output) and not (64 <= output[end] <= 126):
+                                end += 1
+                            end += 1  # include final letter
+                            chunks.append(output[i:end])
+                            i = end
+                        else:
+                            j = i
+                            while j < len(output) and output[j:j+1] != b'\x1b':
+                                j += 1
+                            # Perform replacement on plain text only
+                            text = output[i:j].replace(b"fizz", b"fizzbuzz")
+                            chunks.append(text)
+                            i = j
+
+                    os.write(sys.stdout.fileno(), b''.join(chunks))
+
+    finally:
+        termios.tcsetattr(sys.stdin.fileno(), termios.TCSADRAIN, orig_attrs)
+
+if __name__ == "__main__":
+    main()
+

{streamdown-0.17.0 → streamdown-0.19.0}/streamdown/sd.py

@@ -36,7 +36,7 @@ from functools import reduce
 from argparse import ArgumentParser
 from pygments import highlight
 from pygments.lexers import get_lexer_by_name
-from pygments.formatters import Terminal256Formatter
+from pygments.formatters import TerminalTrueColorFormatter
 from pygments.styles import get_style_by_name
 
 if __package__ is None:
@@ -46,7 +46,7 @@ else:
 
 default_toml = """
 [features]
-CodeSpaces = true
+CodeSpaces = false
 Clipboard  = true
 Logging    = false
 Timeout    = 0.1
@@ -62,7 +62,7 @@ HSV     = [0.8, 0.5, 0.5]
 Dark    = { H = 1.00, S = 1.50, V = 0.25 }
 Mid     = { H = 1.00, S = 1.00, V = 0.50 }
 Symbol  = { H = 1.00, S = 1.00, V = 1.50 }
-Head    = { H = 1.00, S = 2.00, V = 1.50 }
+Head    = { H = 1.00, S = 1.00, V = 1.75 }
 Grey    = { H = 1.00, S = 0.25, V = 1.37 }
 Bright  = { H = 1.00, S = 2.00, V = 2.00 }
 Syntax  = "monokai"
@@ -99,7 +99,8 @@ ANSIESCAPE = r'\033(?:\[[0-9;?]*[a-zA-Z]|][0-9]*;;.*?\\|\\)'
 KEYCODE_RE = re.compile(r'\x1B(?:[@-Z\\-_]|\[[0-?]*[ -/]*[@-~])')
 
 visible = lambda x: re.sub(ANSIESCAPE, "", x)
-visible_length = lambda x: len(visible(x))
+# cjk characters are double width
+visible_length = lambda x: len(visible(x)) + cjk_count(x)
 extract_ansi_codes = lambda text: re.findall(ESCAPE, text)
 remove_ansi = lambda line, codeList: reduce(lambda line, code: line.replace(code, ''), codeList, line)
 
@@ -231,7 +232,7 @@ def format_table(rowList):
     # Note this is where every cell is formatted so if 
     # you are styling, do it before here!
     for row in rowList:
-        wrapped_cell = text_wrap(row, width=col_width)
+        wrapped_cell = text_wrap(row, width=col_width, force_truncate=True)
 
         # Ensure at least one line, even for empty cells
         if not wrapped_cell:
@@ -285,7 +286,7 @@ def code_wrap(text_in):
     # get the indentation of the first line
     indent = len(text_in) - len(text_in.lstrip())
     text = text_in.lstrip()
-    mywidth = state.full_width() - indent
+    mywidth = state.full_width(-4 if Style.PrettyBroken else 0) - indent
 
     # We take special care to preserve empty lines
     if len(text) == 0:
@@ -304,7 +305,7 @@ def ansi_collapse(codelist, inp):
     # We break SGR strings into various classes concerning their applicate or removal
     nums = {
         'fg': r'3\d', 'bg': r'4\d',
-        'b': r'2?1', 'i': r'2?3', 'u': r'3?2',
+        'b': r'2?[12]', 'i': r'2?3', 'u': r'3?2',
         'reset': '0'
     }
 
@@ -334,16 +335,24 @@ def ansi_collapse(codelist, inp):
 
     return codelist + inp
 
-def text_wrap(text, width = -1, indent = 0, first_line_prefix="", subsequent_line_prefix=""):
+
+def split_text(text):
+    return [x for x in re.split(
+        r'(?<=[\u3000-\u303F\u4E00-\u9FFF\u3400-\u4DBF\uF900-\uFAFF])|(?=[\u4E00-\u9FFF\u3400-\u4DBF\uF900-\uFAFF])|\s+',
+        text
+    ) if x]
+
+def text_wrap(text, width = -1, indent = 0, first_line_prefix="", subsequent_line_prefix="", force_truncate=False):
     if width == -1:
         width = state.Width
 
     # The empty word clears the buffer at the end.
-    words = line_format(text).split() + [""]
+    words = split_text(line_format(text)) + [""]
     lines = []
     current_line = ""
     current_style = []
     
+    oldword = ''
     for word in words:
         # we apply the style if we see it at the beginning of the word
         codes = extract_ansi_codes(word)
@@ -352,13 +361,25 @@ def text_wrap(text, width = -1, indent = 0, first_line_prefix="", subsequent_lin
             current_style.append(codes.pop(0))
 
         if len(word) and visible_length(current_line) + visible_length(word) + 1 <= width:  # +1 for space
-            current_line += (" " if current_line else "") + word
+            space = ""
+            if len(visible(word)) > 0 and current_line:
+                space = " "
+            if (":" in visible(word) or cjk_count(word)) and cjk_count(oldword):
+                space = ""
+            current_line += space + word
         else:
             # Word doesn't fit, finalize the previous line
             prefix = first_line_prefix if not lines else subsequent_line_prefix
             line_content = prefix + current_line
+            # This is expensive, fix.
+            while force_truncate and visible_length(line_content) >= width:
+                line_content = line_content[:len(line_content) - 2] + "…"
+
             margin = max(0, width - visible_length(line_content))
-            lines.append(line_content + state.bg + ' ' * margin)
+
+            if line_content.strip() != "":
+                lines.append(line_content + state.bg + ' ' * margin)
+
             current_line = (" " * indent) + "".join(current_style) + word
 
         if len(codes):
@@ -367,13 +388,27 @@ def text_wrap(text, width = -1, indent = 0, first_line_prefix="", subsequent_lin
         if codes:
             current_style = ansi_collapse(current_style, codes)
 
+        oldword = word
+
     if len(lines) < 1:
         return []
 
     return lines
 
+def cjk_count(s):
+    cjk_re = re.compile(
+        r'[\u4E00-\u9FFF'      # CJK Unified Ideographs
+        r'|\u3400-\u4DBF'       # CJK Unified Ideographs Extension A
+        r'|\uF900-\uFAFF'       # CJK Compatibility Ideographs
+		r'|\uFF00-\uFFEF'       # CJK Compatibility Punctuation
+        r'|\u3000-\u303F'      # CJK Symbols and Punctuation
+        r'|\U0002F800-\U0002FA1F]' # CJK Compatibility Ideographs Supplement
+    )
+    
+    return len(cjk_re.findall(visible(s)))
+
 def line_format(line):
-    not_text = lambda token: not (token.isalnum() or token == '\\')
+    not_text = lambda token: not (token.isalnum() or token == '\\') or cjk_count(token)
     footnotes = lambda match: ''.join([chr(SUPER[int(i)]) for i in match.group(1)])
 
     def process_images(match):
@@ -402,7 +437,7 @@ def line_format(line):
     result = ""
 
     for match in tokenList:
-        token = match.group(1)
+        token = re.sub(r'\s+',' ', match.group(1))
         next_token = line[match.end()] if match.end() < len(line) else ""
         prev_token = line[match.start()-1] if match.start() > 0 else ""
 
@@ -458,6 +493,7 @@ def parse(stream):
     last_line_empty_cache = None
     byte = None
     TimeoutIx = 0
+    lexer = None
     while True:
         if state.is_pty or state.is_exec:
             byte = None
@@ -548,7 +584,7 @@ def parse(stream):
                 line = line[len(block_match.group(0)):]
         else:
             if state.block_depth > 0:
-                line = FGRESET + line
+                yield FGRESET
                 state.block_depth = 0
 
         # --- Collapse Multiple Empty Lines if not in code blocks ---
@@ -569,7 +605,7 @@ def parse(stream):
         # \n buffer
         if not state.in_list and len(state.ordered_list_numbers) > 0:
             state.ordered_list_numbers[0] = 0
-        elif not line.startswith(' ' * state.list_indent_text):
+        elif (not line.startswith(' ' * state.list_indent_text)) and line.strip() != "":
             state.in_list = False
             state.list_indent_text = 0
 
@@ -578,7 +614,7 @@ def parse(stream):
         if len(line) - len(line.lstrip()) >= state.first_indent:
             line = line[state.first_indent:]
         else:
-            logging.warning("Indentation decreased from first line.")
+            logging.debug("Indentation decreased from first line.")
 
 
         # Indent guaranteed
@@ -648,8 +684,8 @@ def parse(stream):
                     logging.debug(f"code: {state.in_code}")
                     state.emit_flush = True
                     # We suppress the newline - it's not an explicit style
-                    state.has_newline = False
-                    yield RESET
+                    #state.has_newline = False
+                    #yield RESET
 
                     if code_type == Code.Backtick:
                         continue
@@ -658,7 +694,7 @@ def parse(stream):
                         # nor do we want to be here.
                         raise Goto()
 
-                if state.code_first_line:
+                if state.code_first_line or lexer is None:
                     state.code_first_line = False
                     try:
                         lexer = get_lexer_by_name(state.code_language)
@@ -667,7 +703,7 @@ def parse(stream):
                         lexer = get_lexer_by_name("Bash")
                         custom_style = get_style_by_name("default")
 
-                    formatter = Terminal256Formatter(style=custom_style)
+                    formatter = TerminalTrueColorFormatter(style=custom_style)
                     if line.startswith(' ' * state.code_indent):
                         line = line[state.code_indent :]
 
@@ -687,7 +723,7 @@ def parse(stream):
                 indent, line_wrap = code_wrap(line)
                 
                 state.where_from = "in code"
-                pre = [state.space_left(listwidth = True), ' '] if Style.PrettyBroken else ['', '']
+                pre = [state.space_left(listwidth = True), '  '] if Style.PrettyBroken else ['', '']
 
                 for tline in line_wrap:
                     # wrap-around is a bunch of tricks. We essentially format longer and longer portions of code. The problem is
@@ -698,7 +734,7 @@ def parse(stream):
 
                     # Sometimes the highlighter will do things like a full reset or a background reset.
                     # This is not what we want
-                    highlighted_code = re.sub(r"\033\[39(;00|)m", '', highlighted_code)
+                    highlighted_code = re.sub(r"\033\[49(;00|)m", '', highlighted_code)
     
                     # Since we are streaming we ignore the resets and newlines at the end
                     if highlighted_code.endswith(FGRESET + "\n"):
@@ -805,7 +841,7 @@ def parse(stream):
         # This is intentional ... we can get here in llama 4 using
         # a weird thing
         if state.in_list:
-            indent = (len(state.list_item_stack) - 1) * Style.ListIndent
+            indent = (len(state.list_item_stack) - 1) * Style.ListIndent #+ (len(bullet) - 1)
             wrap_width = state.current_width() - indent - (2 * Style.ListIndent) 
             
             wrapped_lineList = text_wrap(content, wrap_width, Style.ListIndent,
@@ -964,6 +1000,7 @@ def main():
             os.close(state.exec_slave)  # We don't need slave in parent
             # Set stdin to raw mode so we don't need to press enter
             tty.setcbreak(sys.stdin.fileno())
+            sys.stdout.write("\x1b[?7h")
             emit(sys.stdin)
 
         elif args.filenameList:

streamdown-0.19.0/streamdown/ss

@@ -0,0 +1 @@
+*   **Model Card:**  Always read the model card on the Hugging Face Hub ([https://huggingface.co/microsoft/bitnet-b1.58-2B-4T](https://huggingface.co/microsoft/bitnet-b1.58-2B-4T)) for important information about the model, its intended use, limitations, and potential biases.

streamdown-0.19.0/streamdown/ss1

@@ -0,0 +1,42 @@
+    *   `model.safetensors`: The name of the file you want to download.  You'll need to know the exact filename.  You can find the files in the model repository on the Hugging Face Hub website ([https://huggingface.co/microsoft/bitnet-b1.58-2B-4T](https://huggingface.co/microsoft/bitnet-b1.58-2B-4T)).  Look under the "Files and versions" tab.  `safetensors` is the preferred format for model weights now.  If it's a `.bin` file, you can download that instead.
+    *   `--local-dir ./bitnet-b1.58-2B-4T`:  The directory to save the file to.
+
+*   **Download using `transformers` library (recommended for most use cases):**
+
+    The `transformers` library provides a convenient way to download and cache models.  This is often the easiest approach if you're using the model with `transformers`.  You don't *directly* use the `huggingface-cli` for this, but it's worth knowing.
+
+    ```python
+    from transformers import AutoModelForCausalLM, AutoTokenizer
+
+    model_name = "microsoft/bitnet-b1.58-2B-4T"
+
+    tokenizer = AutoTokenizer.from_pretrained(model_name)
+    model = AutoModelForCausalLM.from_pretrained(model_name)
+
+    # The model and tokenizer will be downloaded and cached in your
+    # transformers cache directory (usually ~/.cache/huggingface/transformers).
+    ```
+
+    This approach automatically handles downloading the necessary files and caching them for future use.  It also handles the correct file formats and configurations.
+
+**4.  Checking the Download**
+
+After the download completes, verify that the files are in the specified directory.  You can use `ls` (Linux/macOS) or `dir` (Windows) to list the contents of the directory.
+
+**Important Considerations:**
+
+*   **Disk Space:**  The `bitnet-b1.58-2B-4T` model is quite large (several gigabytes).  Make sure you have enough free disk space before downloading.
+*   **Network Connection:**  A stable and fast internet connection is essential for a smooth download.
+*   **Caching:** The Hugging Face Hub and `transformers` library use caching to avoid re-downloading models unnecessarily.  The default cache directory is usually `~/.cache/huggingface/transformers`.
+*   **File Formats:**  Models are often stored in `safetensors` or `.bin` formats.  `safetensors` is generally preferred for security and performance.
+*   **Model Card:**  Always read the model card on the Hugging Face Hub ([https://huggingface.co/microsoft/bitnet-b1.58-2B-4T](https://huggingface.co/microsoft/bitnet-b1.58-2B-4T)) for important information about the model, its intended use, limitations, and potential biases.
+*   **Gated Models:** Some models require you to accept terms of use before you can download them.  The `huggingface-cli login` command will guide you through this process if necessary.
+
+**Example Workflow (Recommended):**
+
+1.  `huggingface-cli login` (if not already logged in)
+2.  Use the `transformers` library in a Python script to download and load the model (as shown in the example above).  This is the most convenient and reliable method for most use cases.
+
+Let me know if you have any other questions or if you'd like help with a specific task related to this model!
+
+> 

streamdown-0.19.0/tests/bold_reset_with_link.md

@@ -0,0 +1 @@
+*   **Model Card:**  When this wraps with this link ([https://huggingface.co/microsoft/bitnet-b1.58-2B-4T](https://huggingface.co/microsoft/bitnet-b1.58-2B-4T)) the next line used to be bold because of some stripping error. This should no longer be the case

streamdown-0.19.0/tests/chinese.md

@@ -0,0 +1,4 @@
+**XIǍO LÓNG:** *(mocking the academics)*  
+"Ohhh, ‘**Dadaist**’? ‘**Absurd**’? 那你们怎是在说**模形式**？你们这群人连**京剧密码**都解么没看懂那个**转身**是在问**递归关系**？那个**抖袖**是在说**模形式**？你们这群人连**京剧密码**都解不开，还好意思说自己是数学家？"
+*(Translation: "Ohhh, ‘Dadaist’? ‘Absurd’? Then how come none of you realized the **spin** was about **recurrence relations**? The **sleeve flick** was about **modular forms**? You can’t even crack **Peking Opera code**, and you call yourselves mathematicians?")*
+

streamdown-0.19.0/tests/cjj.mv

@@ -0,0 +1,4 @@
+| 特性               | Ruby                            | Python                          | OCaml                          |
+|------------------|--------------------------------|--------------------------------|-------------------------------|
+| **适用场景**       | 快速开发、Web 应用、自动化脚本    | 数据科学、机器学习、Web 开发、自动化脚本 | 性能关键的应用、编译器开发、金融建模 |
+

streamdown-0.19.0/tests/cjk-table.md

@@ -0,0 +1,13 @@
+| 特性               | Ruby                            | Python                          | OCaml                          |
+|------------------|--------------------------------|--------------------------------|-------------------------------|
+| **设计哲学**       | 使编程更加自然和有趣              | 强调代码的可读性和简洁性          | 提供高效的函数式编程功能        |
+| **应用领域**       | Web 开发、自动化脚本、原型开发    | Web 开发、数据科学、机器学习、自动化脚本 | 编译器开发、系统编程、金融建模  |
+| **特点**           | 动态类型、面向对象、元编程能力强  | 动态类型、面向对象、大量第三方库支持  | 静态类型、函数式编程、模块化、高性能 |
+| **语法示例**       | ```ruby<br>class Greeter<br>  def initialize(name)<br>    @name = name<br>  end<br><br>  def greet<br>    puts "Hello, #{@name}!"<br>  end<br>end<br><br>greeter = Greeter.new("World")<br>greeter.greet``` | ```python<br>class Greeter:<br>  def __init__(self, name):<br>    self.name = name<br><br>  def greet(self):<br>    print(f"Hello, {self.name}!")<br><br>greeter = Greeter("World")<br>greeter.greet()``` | ```ocaml<br>type greeter = {<br>  name: string;<br>}<br><br>let greet {name} =<br>  print_endline ("Hello, " ^ name ^ "!")<br><br>let greeter = {name = "World"}<br>let () = greet greeter``` |
+| **性能**           | 相对较低，但通过 JRuby、TruffleRuby 可以提高性能 | 适中，通过 Cython、PyPy 可以显著提高性能 | 高，编译后的代码接近 C 语言的性能 |
+| **优点**           | 开发速度快，社区活跃             | 丰富的第三方库，应用广泛          | 适合性能关键的应用，编译器优化能力强 |
+| **生态系统**       | Ruby on Rails 是最著名的 Web 框架，其他库和工具也很丰富 | 丰富的第三方库，如 NumPy、Pandas、Django、Flask | 相对较少，但有高质量的库，如 Core 和 Async |
+| **学习曲线**       | 较平缓，语法灵活，适合初学者      | 平坦，语法清晰，适合初学者        | 较陡峭，需要理解函数式编程概念    |
+| **文档**           | 丰富，社区支持好                 | 丰富，有大量的教程和资源          | 适中，但社区提供的资源质量高      |
+| **适用场景**       | 快速开发、Web 应用、自动化脚本    | 数据科学、机器学习、Web 开发、自动化脚本 | 性能关键的应用、编译器开发、金融建模 |
+