23224625495

Committed 18 Mar 2026 01:26AM UTC coverage: 37.013%. First build

Build # 23224625495

Build Type

push

github

Committed by

web-flow

Commit Message

Refactor caching, examples, and bins (#34)

Coverage Stats

273 of 900 new or added lines in 37 files covered. (30.33%)

1244 of 3361 relevant lines covered (37.01%)

1707.77 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

0.0

/src/models/filing_document.rs

/// Represents a single document entry within an SEC EDGAR filing index.
///
/// Parsed from the EDGAR HTML filing index page (`{accession}-index.htm`).
///
/// # Common `document_type` values
///
/// | Type pattern | Description |
/// |---|---|
/// | `"8-K"`, `"10-K"`, `"10-Q"` | Primary form body |
/// | `"EX-99.1"`, `"EX-99.2"` | Press release / earnings release |
/// | `"EX-10.1"` … `"EX-10.x"` | Material contracts |
/// | `"EX-21.1"` | Subsidiaries of the registrant |
/// | `"EX-23.1"` | Auditor / accountant consent |
/// | `"EX-31.1"`, `"EX-31.2"` | SOX § 302 certifications (CEO/CFO) |
/// | `"EX-32.1"`, `"EX-32.2"` | SOX § 906 certifications (CEO/CFO) |
/// | `"EX-101.INS"` … `"EX-101.PRE"` | XBRL instance / schema / label files |
/// | `"GRAPHIC"` | Image files (logos, charts) |
/// | `"XML"`, `""` | Ancillary machine-readable files |
#[derive(Debug, Clone)]
pub struct FilingDocument {
    /// File name within the filing archive (e.g. `"ex991.htm"`).
    pub name: String,

    /// SEC document type string (e.g. `"EX-99.1"`, `"8-K"`).
    pub document_type: String,
}

impl FilingDocument {
    /// Returns `true` if this document is an exhibit (type starts with `"EX-"`).
    pub fn is_exhibit(&self) -> bool {
        self.document_type.to_uppercase().starts_with("EX-")
    }

    /// Returns `true` if this is a press release or earnings release exhibit.
    ///
    /// Matches `EX-99.x` — the standard type used for press releases,
    /// earnings tables, and Regulation FD voluntary disclosures.
    pub fn is_press_release(&self) -> bool {
        let t = self.document_type.to_uppercase();
        t.starts_with("EX-99")
    }

    /// Returns `true` if this is a Sarbanes-Oxley certification exhibit.
    ///
    /// Matches `EX-31.x` (§ 302 CEO/CFO certifications) and `EX-32.x`
    /// (§ 906 certifications).  These are short, formulaic legal documents
    /// with no analytical content.
    pub fn is_sarbanes_oxley_cert(&self) -> bool {
        let t = self.document_type.to_uppercase();
        t.starts_with("EX-31") || t.starts_with("EX-32")
    }

    /// Returns `true` if this is an auditor or accountant consent exhibit (`EX-23.x`).
    ///
    /// Consent forms are short single-page documents required when financial
    /// statements are incorporated by reference; they contain no analytical
    /// content.
    pub fn is_auditor_consent(&self) -> bool {
        self.document_type.to_uppercase().starts_with("EX-23")
    }

    /// Returns `true` if this is an XBRL structured data exhibit (`EX-101.*`).
    ///
    /// These are machine-readable schema, instance, label, and calculation
    /// documents that duplicate the financial data already present in the
    /// primary HTML document.  They contain no human-readable prose.
    pub fn is_xbrl_data(&self) -> bool {
        self.document_type.to_uppercase().starts_with("EX-101")
    }

    /// Returns `true` if this exhibit contains substantive, human-readable content.
    ///
    /// An exhibit is substantive if it:
    /// - Is an `EX-*` document, **and**
    /// - Is **not** a SOX certification (`EX-31.x`, `EX-32.x`)
    /// - Is **not** an auditor consent (`EX-23.x`)
    /// - Is **not** an XBRL data file (`EX-101.*`)
    /// - Is **not** a raw graphic (`GRAPHIC`)
    ///
    /// Substantive exhibits include press releases (`EX-99.x`), material
    /// contracts (`EX-10.x`), subsidiary lists (`EX-21.x`), and any other
    /// exhibit that contains actual prose or financial tables.
    ///
    /// Use [`FilingIndex::substantive_exhibits`] to filter a filing's documents
    /// to only those worth rendering or embedding.
    pub fn is_substantive_exhibit(&self) -> bool {
        if !self.is_exhibit() {
            return false;
        }
        if self.is_sarbanes_oxley_cert() || self.is_auditor_consent() || self.is_xbrl_data() {
            return false;
        }
        if self.document_type.to_uppercase() == "GRAPHIC" {
            return false;
        }
        true
    }

    /// Returns `true` if the file is an HTML document.
    pub fn is_html(&self) -> bool {
        let n = self.name.to_lowercase();
        n.ends_with(".htm") || n.ends_with(".html")
    }

    /// Returns `true` if the file is a plain-text document.
    pub fn is_text(&self) -> bool {
        self.name.to_lowercase().ends_with(".txt")
    }
}

/// The full document listing for a single SEC EDGAR filing.
///
/// Fetch via [`crate::network::fetch_filing_index`].
#[derive(Debug)]
pub struct FilingIndex {
    /// All documents listed in this filing archive.
    pub documents: Vec<FilingDocument>,
}

impl FilingIndex {
    /// Returns all exhibit documents from this filing (any `EX-*` type).
    ///
    /// Includes everything: press releases, SOX certs, auditor consents,
    /// XBRL data files, and graphics.  Use [`Self::substantive_exhibits`]
    /// to filter to only human-readable content.
    pub fn exhibits(&self) -> Vec<&FilingDocument> {
        self.documents
            .iter()
            .filter(|doc| doc.is_exhibit())
            .collect()
    }

    /// Returns only the substantive exhibits — those containing human-readable
    /// prose or financial tables.
    ///
    /// Excludes SOX certifications, auditor consents, XBRL data files, and
    /// graphics.  This is the right default filter when rendering or embedding
    /// a filing's attachments.
    ///
    /// See [`FilingDocument::is_substantive_exhibit`] for the exact criteria.
    pub fn substantive_exhibits(&self) -> Vec<&FilingDocument> {
        self.documents
            .iter()
            .filter(|doc| doc.is_substantive_exhibit())
            .collect()
    }

    /// Returns only press release exhibits (`EX-99.x`).
    ///
    /// These are typically earnings announcements, Regulation FD voluntary
    /// disclosures, and other company-authored statements attached to 8-K
    /// filings.  To find earnings releases specifically, check the parent
    /// [`CikSubmission::is_earnings_release`] before fetching the index.
    ///
    /// [`CikSubmission::is_earnings_release`]: crate::models::CikSubmission::is_earnings_release
    pub fn press_releases(&self) -> Vec<&FilingDocument> {
        self.documents
            .iter()
            .filter(|doc| doc.is_press_release())
            .collect()
    }
}

1	/// Represents a single document entry within an SEC EDGAR filing index.
2	///
3	/// Parsed from the EDGAR HTML filing index page (`{accession}-index.htm`).
4	///
5	/// # Common `document_type` values
6	///
7	/// \| Type pattern \| Description \|
8	/// \|---\|---\|
9	/// \| `"8-K"`, `"10-K"`, `"10-Q"` \| Primary form body \|
10	/// \| `"EX-99.1"`, `"EX-99.2"` \| Press release / earnings release \|
11	/// \| `"EX-10.1"` … `"EX-10.x"` \| Material contracts \|
12	/// \| `"EX-21.1"` \| Subsidiaries of the registrant \|
13	/// \| `"EX-23.1"` \| Auditor / accountant consent \|
14	/// \| `"EX-31.1"`, `"EX-31.2"` \| SOX § 302 certifications (CEO/CFO) \|
15	/// \| `"EX-32.1"`, `"EX-32.2"` \| SOX § 906 certifications (CEO/CFO) \|
16	/// \| `"EX-101.INS"` … `"EX-101.PRE"` \| XBRL instance / schema / label files \|
17	/// \| `"GRAPHIC"` \| Image files (logos, charts) \|
18	/// \| `"XML"`, `""` \| Ancillary machine-readable files \|
19	#[derive(Debug, Clone)]
20	pub struct FilingDocument {
21	/// File name within the filing archive (e.g. `"ex991.htm"`).
22	pub name: String,
23
24	/// SEC document type string (e.g. `"EX-99.1"`, `"8-K"`).
25	pub document_type: String,
26	}
27
28	impl FilingDocument {
29	/// Returns `true` if this document is an exhibit (type starts with `"EX-"`).
30	pub fn is_exhibit(&self) -> bool {	×
31	self.document_type.to_uppercase().starts_with("EX-")	×
32	}	×
33
34	/// Returns `true` if this is a press release or earnings release exhibit.
35	///
36	/// Matches `EX-99.x` — the standard type used for press releases,
37	/// earnings tables, and Regulation FD voluntary disclosures.
NEW 38	pub fn is_press_release(&self) -> bool {	×
NEW 39	let t = self.document_type.to_uppercase();	×
NEW 40	t.starts_with("EX-99")	×
NEW 41	}	×
42
43	/// Returns `true` if this is a Sarbanes-Oxley certification exhibit.
44	///
45	/// Matches `EX-31.x` (§ 302 CEO/CFO certifications) and `EX-32.x`
46	/// (§ 906 certifications). These are short, formulaic legal documents
47	/// with no analytical content.
NEW 48	pub fn is_sarbanes_oxley_cert(&self) -> bool {	×
NEW 49	let t = self.document_type.to_uppercase();	×
NEW 50	t.starts_with("EX-31") \|\| t.starts_with("EX-32")	×
NEW 51	}	×
52
53	/// Returns `true` if this is an auditor or accountant consent exhibit (`EX-23.x`).
54	///
55	/// Consent forms are short single-page documents required when financial
56	/// statements are incorporated by reference; they contain no analytical
57	/// content.
NEW 58	pub fn is_auditor_consent(&self) -> bool {	×
NEW 59	self.document_type.to_uppercase().starts_with("EX-23")	×
NEW 60	}	×
61
62	/// Returns `true` if this is an XBRL structured data exhibit (`EX-101.*`).
63	///
64	/// These are machine-readable schema, instance, label, and calculation
65	/// documents that duplicate the financial data already present in the
66	/// primary HTML document. They contain no human-readable prose.
NEW 67	pub fn is_xbrl_data(&self) -> bool {	×
NEW 68	self.document_type.to_uppercase().starts_with("EX-101")	×
NEW 69	}	×
70
71	/// Returns `true` if this exhibit contains substantive, human-readable content.
72	///
73	/// An exhibit is substantive if it:
74	/// - Is an `EX-` document, and*
75	/// - Is not a SOX certification (`EX-31.x`, `EX-32.x`)
76	/// - Is not an auditor consent (`EX-23.x`)
77	/// - Is not an XBRL data file (`EX-101.*`)
78	/// - Is not a raw graphic (`GRAPHIC`)
79	///
80	/// Substantive exhibits include press releases (`EX-99.x`), material
81	/// contracts (`EX-10.x`), subsidiary lists (`EX-21.x`), and any other
82	/// exhibit that contains actual prose or financial tables.
83	///
84	/// Use [`FilingIndex::substantive_exhibits`] to filter a filing's documents
85	/// to only those worth rendering or embedding.
NEW 86	pub fn is_substantive_exhibit(&self) -> bool {	×
NEW 87	if !self.is_exhibit() {	×
NEW 88	return false;	×
NEW 89	}	×
NEW 90	if self.is_sarbanes_oxley_cert() \|\| self.is_auditor_consent() \|\| self.is_xbrl_data() {	×
NEW 91	return false;	×
NEW 92	}	×
NEW 93	if self.document_type.to_uppercase() == "GRAPHIC" {	×
NEW 94	return false;	×
NEW 95	}	×
NEW 96	true	×
NEW 97	}	×
98
99	/// Returns `true` if the file is an HTML document.
100	pub fn is_html(&self) -> bool {	×
101	let n = self.name.to_lowercase();	×
102	n.ends_with(".htm") \|\| n.ends_with(".html")	×
103	}	×
104
105	/// Returns `true` if the file is a plain-text document.
106	pub fn is_text(&self) -> bool {	×
107	self.name.to_lowercase().ends_with(".txt")	×
108	}	×
109	}
110
111	/// The full document listing for a single SEC EDGAR filing.
112	///
113	/// Fetch via [`crate::network::fetch_filing_index`].
114	#[derive(Debug)]
115	pub struct FilingIndex {
116	/// All documents listed in this filing archive.
117	pub documents: Vec<FilingDocument>,
118	}
119
120	impl FilingIndex {
121	/// Returns all exhibit documents from this filing (any `EX-*` type).
122	///
123	/// Includes everything: press releases, SOX certs, auditor consents,
124	/// XBRL data files, and graphics. Use [`Self::substantive_exhibits`]
125	/// to filter to only human-readable content.
126	pub fn exhibits(&self) -> Vec<&FilingDocument> {	×
127	self.documents	×
128	.iter()	×
129	.filter(\|doc\| doc.is_exhibit())	×
130	.collect()	×
131	}	×
132
133	/// Returns only the substantive exhibits — those containing human-readable
134	/// prose or financial tables.
135	///
136	/// Excludes SOX certifications, auditor consents, XBRL data files, and
137	/// graphics. This is the right default filter when rendering or embedding
138	/// a filing's attachments.
139	///
140	/// See [`FilingDocument::is_substantive_exhibit`] for the exact criteria.
NEW 141	pub fn substantive_exhibits(&self) -> Vec<&FilingDocument> {	×
NEW 142	self.documents	×
NEW 143	.iter()	×
NEW 144	.filter(\|doc\| doc.is_substantive_exhibit())	×
NEW 145	.collect()	×
NEW 146	}	×
147
148	/// Returns only press release exhibits (`EX-99.x`).
149	///
150	/// These are typically earnings announcements, Regulation FD voluntary
151	/// disclosures, and other company-authored statements attached to 8-K
152	/// filings. To find earnings releases specifically, check the parent
153	/// [`CikSubmission::is_earnings_release`] before fetching the index.
154	///
155	/// [`CikSubmission::is_earnings_release`]: crate::models::CikSubmission::is_earnings_release
NEW 156	pub fn press_releases(&self) -> Vec<&FilingDocument> {	×
NEW 157	self.documents	×
NEW 158	.iter()	×
NEW 159	.filter(\|doc\| doc.is_press_release())	×
NEW 160	.collect()	×
NEW 161	}	×
162	}

jzombie / rust-sec-fetcher / 23224625495

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous