Coverage for src / lilbee / cli / tui / widgets / crawl_dialog.py: 100%

1"""Modal dialog for configuring a web crawl.""" 

2 

3from __future__ import annotations 

4 

5from dataclasses import dataclass 

6from typing import ClassVar 

7 

8from textual.app import ComposeResult 

9from textual.binding import Binding, BindingType 

10from textual.containers import Center, Vertical 

11from textual.screen import ModalScreen 

12from textual.widgets import Button, Checkbox, Collapsible, Input, Label, Static 

13 

14from lilbee.cli.tui import messages as msg 

15 

16 

17@dataclass(frozen=True) 

18class CrawlParams: 

19 """Validated crawl parameters returned by CrawlDialog. 

20 

21 depth: None = whole-site unbounded. 0 = single URL only. Positive int = 

22 explicit link-follow depth cap. max_pages: None = no cap. Positive int = 

23 explicit page cap. 

24 """ 

25 

26 url: str 

27 depth: int | None 

28 max_pages: int | None 

29 

30 

31class CrawlDialog(ModalScreen[CrawlParams | None]): 

32 """Modal dialog that collects URL, recursion toggle, and optional caps.""" 

33 

34 CSS_PATH = "crawl_dialog.tcss" 

35 AUTO_FOCUS = "#crawl-url-input" 

36 

37 BINDINGS: ClassVar[list[BindingType]] = [ 

38 Binding("escape", "cancel", "Cancel", show=False), 

39 ] 

40 

41 def compose(self) -> ComposeResult: 

42 with Vertical(): 

43 yield Static(msg.CRAWL_DIALOG_TITLE, id="crawl-title") 

44 yield Label(msg.CRAWL_DIALOG_URL_LABEL) 

45 yield Input( 

46 placeholder=msg.CRAWL_DIALOG_URL_PLACEHOLDER, 

47 id="crawl-url-input", 

48 ) 

49 yield Checkbox( 

50 msg.CRAWL_DIALOG_RECURSIVE_LABEL, 

51 value=True, 

52 id="crawl-recursive-checkbox", 

53 ) 

54 with Collapsible(title=msg.CRAWL_DIALOG_ADVANCED_TITLE, id="crawl-advanced"): 

55 yield Label(msg.CRAWL_DIALOG_DEPTH_LABEL, classes="crawl-field-label") 

56 yield Input( 

57 placeholder=msg.CRAWL_DIALOG_DEPTH_PLACEHOLDER, 

58 id="crawl-depth-input", 

59 ) 

60 yield Label(msg.CRAWL_DIALOG_MAX_PAGES_LABEL, classes="crawl-field-label") 

61 yield Input( 

62 placeholder=msg.CRAWL_DIALOG_MAX_PAGES_PLACEHOLDER, 

63 id="crawl-max-pages-input", 

64 ) 

65 yield Static("", id="crawl-error") 

66 with Center(): 

67 yield Button(msg.CRAWL_DIALOG_SUBMIT, variant="primary", id="crawl-submit") 

68 yield Button(msg.CRAWL_DIALOG_CANCEL, variant="default", id="crawl-cancel") 

69 

70 def on_button_pressed(self, event: Button.Pressed) -> None: 

71 if event.button.id == "crawl-submit": 

72 self._try_submit() 

73 else: 

74 self.dismiss(None) 

75 

76 def on_input_submitted(self, _event: Input.Submitted) -> None: 

77 self._try_submit() 

78 

79 @staticmethod 

80 def _parse_optional_non_negative_int(value: str) -> int | None: 

81 """Parse a non-negative integer from *value*; empty string returns None. 

82 

83 None means "no cap" in the crawl API. Zero is meaningful for the 

84 depth field (single-URL crawl per the crawler contract). Raises 

85 ValueError on non-numeric input or negative integers. 

86 """ 

87 if not value: 

88 return None 

89 n = int(value) 

90 if n < 0: 

91 raise ValueError 

92 return n 

93 

94 @staticmethod 

95 def _parse_optional_positive_int(value: str) -> int | None: 

96 """Parse a positive integer from *value*; empty string returns None. 

97 

98 None means "no cap" in the crawl API. Raises ValueError on non-numeric 

99 input or non-positive integers. Used for fields like max_pages where 

100 zero has no useful meaning. 

101 """ 

102 if not value: 

103 return None 

104 n = int(value) 

105 if n <= 0: 

106 raise ValueError 

107 return n 

108 

109 def _validate(self) -> CrawlParams | str: 

110 """Validate inputs. Returns CrawlParams on success, error message on failure.""" 

111 from lilbee.crawler import is_url, require_valid_crawl_url 

112 

113 url = self.query_one("#crawl-url-input", Input).value.strip() 

114 recursive = self.query_one("#crawl-recursive-checkbox", Checkbox).value 

115 depth_str = self.query_one("#crawl-depth-input", Input).value.strip() 

116 max_pages_str = self.query_one("#crawl-max-pages-input", Input).value.strip() 

117 

118 if not url: 

119 return msg.CRAWL_DIALOG_URL_REQUIRED 

120 

121 if not is_url(url): 

122 url = f"https://{url}" 

123 

124 try: 

125 require_valid_crawl_url(url) 

126 except ValueError as exc: 

127 return msg.CRAWL_DIALOG_INVALID_URL.format(error=exc) 

128 

129 if not recursive: 

130 return CrawlParams(url=url, depth=0, max_pages=None) 

131 

132 try: 

133 # depth=0 means "single URL" per the crawler contract; allow it. 

134 depth = self._parse_optional_non_negative_int(depth_str) 

135 except ValueError: 

136 return msg.CRAWL_DIALOG_INVALID_NUMBER.format(field=msg.CRAWL_DIALOG_DEPTH_LABEL) 

137 

138 try: 

139 max_pages = self._parse_optional_positive_int(max_pages_str) 

140 except ValueError: 

141 return msg.CRAWL_DIALOG_INVALID_NUMBER.format(field=msg.CRAWL_DIALOG_MAX_PAGES_LABEL) 

142 

143 return CrawlParams(url=url, depth=depth, max_pages=max_pages) 

144 

145 def _try_submit(self) -> None: 

146 """Validate inputs and dismiss with CrawlParams or show an error.""" 

147 result = self._validate() 

148 error_widget = self.query_one("#crawl-error", Static) 

149 # _validate returns str (error) or CrawlParams; isinstance disambiguates 

150 if isinstance(result, str): 

151 error_widget.update(result) 

152 return 

153 error_widget.update("") 

154 self.dismiss(result) 

155 

156 def action_cancel(self) -> None: 

157 self.dismiss(None)