I don't know of any good "bible"-like books.
I do have some comments on relatively common bond/loan/preferred equity features that I would not expect to be covered adequately in whatever books you find.
Step coupons are common, especially after restructuring or if the issuer is distressed. The coupon might be close to 0 in the first year and then gradually step up to, say, 12% in a few years - if the issuer still hasn't defaulted. Fixed-coupon step-up is more common than floating (where the spread changes).
Some step coupons are not pre-determined at issuance, but instead the bondholder receives bonus/malus (more or less coupon) when e.g. the issuer's credit rating, ESG rating, various covenants/financial ratios, commodity prices, etc get better or worse.
1 callable bonds - traditionally, the call price used to be fixed. However many bonds issued in the last decade include a make-whole call - instead of fixed price, the remaining cash flows are pv'd at treasury+spread, and that's the call price.
2 puttable bonds (demand obligations, tender option bonds) are very common in Russia and India (both foxed and floating coupon), less so in other markets. Variable Rate Demand Obligations (VRDOs) are puttable and seem to have become popular in the U.S. muni markets.
3 0 floor is common on floater coupons, including inflation-linked. Non-0 floors and any caps are much less common.
Gearing is fairly common, i.e. floaters whose coupon forumula involves $g\times$ index. On muni bonds not subjected to income tax, the gearing approximates the tax saving, i.e. the bondholder might receive $57\% \times$ index, while the comparable alternative would to receive the full index, but pay income tax. In some markets, notably Brazil, gearing is used to compensate the bond holder for the credit risk, e.g. $140\%\times$ CDI - similar how to in most markets index + spread would be used instead. Sometimes gearing is used in structured coupons like $\min(\max(\text{gearing} \times (\text{index}_1 - \text{index}_2) , \text{floor}), \text{cap})$.
4 the option to switch between different indices is not very common. There were some Indonesian instruments (USD-denominated) in which the issuer would decide (and announce in e-mail) whether the next coupon would reset from 3-month LIBOR and pay index + spread in 3 months, or reset from 6-month LIBOR and pay in 6 months.
Pay in kind (PIK) is pretty common in high-yield world. From a practical poit of view, we should note that unlike amortization, the factor on a PIK bond increases with every day of accrual. Sometimes the PIK schedule is set at issuance, and sometimes the issuer has the option to decide how much of the next coupon will be in cash, and how much in PIK. Effectively, instead of paying cash coupon, the issuer borrows more money, giving the bond holder more notional of the bond. In extremely rare cases (Russian IAN), the issuer pays PIK with another instrument.
There are many flavors of dual-currency bonds - e.g., the issuer might have the choice to pay a cash flow in currency 1 or 2, using the exchange rate set at bond issuance, which may differ from te market rate at the time of the cash flow - effectively embedding an FX option.
Reference implementations of examples of such instruments in QuantLib and/or FinancePy might be a fun project for someone.
